JP2022188264A - Transcutaneous analyte sensors and monitors, calibration thereof, and associated methods - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide systems and methods for calibrating an analyte concentration sensor within a biological system, generally using only a signal from the analyte concentration sensor.

SOLUTION: For example, at a steady state, an analyte concentration value within the biological system is known, and it may provide a source for calibration. Similar techniques may be employed with slow-moving averages. Variation forms are disclosed.

SELECTED DRAWING: Figure 1

COPYRIGHT: (C)2023,JPO&INPIT

Description

A system and method for processing sensor data from a continuous analyte sensor and for calibration of the sensor.

Diabetes mellitus is a disease in which the pancreas cannot make enough insulin (Type I or insulin dependent) and/or insulin is ineffective (Type 2 or non-insulin dependent). In diabetic conditions, the victim suffers from hyperglycemia, which can lead to many physiological disorders associated with deterioration of small blood vessels, such as renal failure, skin ulcers, or bleeding into the vitreous humor of the eye. There is Hypoglycemic reaction (low blood sugar) is induced by inadvertent overdose of insulin or after normal administration of insulin or glucose-lowering drugs accompanied by abnormal exercise or inadequate food intake. can be

Traditionally, people with diabetes carry self-monitoring blood glucose (SMBG) monitors, which typically require an uncomfortable finger prick. Due to the lack of comfort and convenience, people with diabetes typically only measure their glucose levels 2-4 times per day. Unfortunately, such time intervals are spread so far apart that a person with diabetes may perceive a hyperglycemic or hypoglycemic condition too late, leading to sometimes dangerous side effects. Glucose levels may alternatively be monitored continuously by a sensor system that includes an on-skin sensor assembly. A sensor system may have a wireless transmitter that transmits measurement data to a receiver, which can process and display information based on the measurements.

In the past, various glucose sensors have been developed for continuously measuring glucose levels. Many implantable glucose sensors suffer from complications within the body and provide only short-term and less-than-accurate sensing of blood glucose. Similarly, transcutaneous sensors have faced problems in accurately sensing and reporting back glucose levels continuously over an extended period of time. Efforts have been made to obtain glycemic data from implantable devices and retrospectively determine glycemic trends for analysis, but these efforts remain limited in determining real-time glycemic information. do not assist Efforts have also been made to acquire blood glucose data from transdermal devices for predictive data analysis, but similar problems arise.

In a continuous glucose monitor (CGM), after the sensor is implanted, it is calibrated, after which it provides substantially continuous sensor data to the sensor electronics. The sensor electronics can transform the sensor data so that estimated analyte values are continuously provided to the user. As used herein, the terms “substantially continuous,” “continuously,” etc. refer to a data stream of individual measurements taken at spaced intervals, which are less than 1 second. from up to, for example, 1, 2, or 5 minutes or more. As the sensor electronics continue to receive sensor data, the sensor can be recalibrated from time to time to account for possible changes (drift) in sensor sensitivity and/or baseline. Sensor sensitivity can refer to the amount of current generated in a sensor by a given amount of measured analyte.

Sensor baseline refers to the signal output by the sensor when no analyte is detected. Over time, sensitivity and baseline change due to a variety of factors, including cell attack or migration of cells to the sensor, which can affect the ability of analytes to reach the sensor.

This background is provided to introduce a brief context for the Summary and Detailed Description that follow. This background is intended to aid in determining the scope of the claimed subject matter, nor is it intended to address the claimed subject matter without any of the above disadvantages or problems. or to be construed as limiting implementations to all-encompassing solutions.

Without limiting the scope of the embodiments represented by the following claims, salient features of systems and methods in accordance with the present principles are briefly discussed. After considering this discussion, and particularly after reading the section entitled "Detailed Description," you will understand how the features of the present embodiments provide the advantages described herein. be.

In a first aspect, a method is provided for calibrating an analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, wherein upon occurrence of a repetitive event, an analyte concentration value in the biological system is known, the method comprising detecting, on a monitoring device, when an analyte concentration value measured by an analyte concentration sensor indwelling within a biological system constitutes a first repetitive event; The measured analyte concentration value at the first repetitive event in which the biological system is detected on the monitoring device or on a device or server operably coupled to the monitoring device, with the known analyte concentration value. and correlating.

Implementations of embodiments and aspects can include one or more of the following. Correlating may include determining a functional relationship between sensor readings and known analyte concentration values. A functional relationship may include a multiplicative constant. Detection may include waiting a predetermined amount of time following entry of an event such as eating or exercising on the monitoring device. The method is detecting the occurrence of a second repetitive event following the correlation, wherein the second repetitive event is different from the first repetitive event; recalibrating the analyte concentration sensor by correlating the sensor readings at the second repetitive event in which is detected to known analyte concentration values. The sensor reading may have a first raw value at the initial calibration and a second raw value at the recalibration, the first and second raw values being different. The method comprises, following the correlation, displaying a graph or table showing currently measured and historical values of the calibrated analyte concentration based at least in part on the correlation; Updating a graphical or tabular display showing currently measured and historical values of the analyte concentration in accordance with the recalibration. An update may change the display of the historical value of the analyte concentration. The method includes determining a difference between a first raw value and a second raw value, comparing a quantity based on the difference to a predetermined standard, and performing sensor calibration based on the comparison. and determining whether is drifting. The method may further include determining a quantitative amount by which the sensor calibration has drifted. The method may further comprise adjusting sensor calibration based on the determined quantitative quantity. The quantity may be a slope between a first raw value and a second raw value. The method may further comprise, if the slope exceeds a predetermined threshold, inhibiting future calibrations on a steady state basis until the slope no longer exceeds the predetermined threshold. The method may further include prompting the user to enter the measurements. The sensor may be a glucose sensor. The method may further include, following correlating, receiving a signal from the sensor and displaying a value corresponding to the received signal, the displayed value being a combination of the received signal and the known analyte. Based on density values. The method may further comprise determining the known analyte concentration value by prompting the user to enter the measured value. The method may further comprise determining known analyte concentration values by accessing population means. Recalibration may be configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby eliminating unexpected spikes in readings. Occurrence is reduced and such recalibration at such times may be caused or configured to occur in any of the described embodiments or aspects.

In a second aspect, a method is provided for compensating for analyte concentration sensor drift in a biological system using only the signal from the analyte concentration sensor, wherein the indwelling analyte concentration sensor is used to measuring a value; determining a first slow moving average of the analyte measurements over a first time period; and referencing calibration of the sensor based at least in part on the first slow moving average. and, following the first determination, determining a second slow moving average of the analyte measurements over a second time period; and adjusting the calibration of the sensor based on the difference between .

Implementations of aspects and embodiments can include one or more of the following. The duration of the first time period may be greater than about 12 hours or greater than about 24 hours. The duration of the first period may be the same as the duration of the second period. The method includes referencing calibration of the sensor based at least in part on the first slow moving average, followed by at least historical values of the calibrated analyte concentration based at least in part on the first slow moving average. and, subsequent to the adjustment, updating the display of the graph or table showing at least historical values of the analyte concentration according to the adjusted calibration. An update may change the display of the historical value of the analyte concentration. The displayed graph or table may further indicate the current measured value of the analyte concentration. Basing the calibration of the sensor based at least in part on the first slow moving average further bases the calibration on a seed value, such as a seed value received from a population average or from a previous session. may contain. The method may further include, following the adjustment, changing the seed value based at least in part on the adjustment. The method may further include varying the seed value based on the difference between the first slow moving average and the second slow moving average.

In a third aspect, a method is provided for compensating for analyte concentration sensor drift in a biological system using only the signal from the analyte concentration sensor, wherein the indwelling analyte concentration sensor is used to determining a first slow moving average of the analyte measurements over a first time period; and basing a first apparent sensitivity of the sensor at least in part on the first slow moving average. determining, following the first determination, a second slow moving average of the analyte measurements over a second time period; determining if the change in apparent sensitivity of the sensor between the first apparent sensitivity and the second apparent sensitivity meets a predetermined criterion; and changing the apparent sensitivity. matches a predetermined criterion, adjusting the actual sensitivity of the sensor to the adjusted value based on the difference between the first apparent sensitivity and the second apparent sensitivity; are not met, prompting the user to enter data from which changes in apparent sensitivity can be determined.

Implementations of aspects and embodiments can include one or more of the following. Determining may include determining whether the apparent sensitivity change is due to sensitivity drift or a slow moving average change. If the apparent sensitivity change is due to a slow moving average change, the method may further include prompting the user to enter data related to the change. Predetermined criteria may include known behavior of sensitivity changes over time for the sensor. The known behavior of sensitivity change may constitute an envelope of acceptable sensitivity change over time. The adjusted value may be based at least in part on the second slow moving average. The predetermined criteria may further include known values for possible physiological analyte changes. The user prompt may include prompting the user to enter calibration values. Prompting the user may include prompting the user to enter diet or exercise information. Upon receiving calibration values or diet or exercise information from the user, the method may further comprise determining whether changes in apparent sensitivity are due to sensitivity drift or slow moving average changes. The method may further comprise adjusting the actual sensitivity of the sensor based on the received calibration value or diet or exercise information.

In a fourth aspect, a method is provided for checking calibration of an analyte concentration sensor system in a biological system using only signals from analyte concentration sensors, wherein after initial calibration, indwelling analyte concentration sensors are used. calculating a clinical value of the analyte concentration based on the measured value and the initial calibration; adjusting the initial calibration to the updated calibration; calculating a clinical value of the analyte concentration based on the measurements and the updated calibration.

Implementations of aspects and embodiments can include one or more of the following. The initial calibration may be based on population averages or data entered by the user. The adjustment may be based on a slow moving average of analyte measurements over time. Adjustment may be based on the steady state value of the analyte. Initial calibration may be based on data determined prior to the session associated with the indwelling analyte concentration sensor. Data may be determined a priori, on the bench, or in vitro.

In a fifth aspect, a method is provided for calibrating an analyte concentration sensor in a biological system using a signal from the analyte concentration sensor, wherein at steady state the analyte concentration value in the biological system is known. , the method comprises receiving, on a monitoring device, a seed value for a calibration parameter; Detecting when and measuring the analyte concentration value when the biological system is at a detected steady state on the monitoring device or on a device or server operably linked to the monitoring device, the known analyte correlating with the concentration values; following the correlation, receiving signals from the sensors; calculating and displaying values corresponding to the received signals, the calculated values being the received signals; calculating and displaying based on known analyte concentration values and seed values.

Implementations of aspects and embodiments can include one or more of the following. The received seed value may be received from a source containing factory calibration information. The method may further include detecting behavior in the received signal other than predetermined parameters and prompting the user to enter external calibration information. The displayed values may be further based on external calibration information. External calibration information may be received from SMBG or fingerstick calibration. The method may further include resetting the known calibration values to new known calibration values, the resetting being based at least in part on the external calibration information. The method may further include resetting the seed value to a new seed value, the reset being based at least in part on the external calibration information. The method may further include altering the display based on the determined accuracy of the value. Display changes may include displaying ranges rather than values, or vice versa. The received seed values for the calibration parameters may be characteristics of the medical condition entered by the user. The user-entered medical condition characteristics may include an indication of type I diabetes, type II diabetes, non-diabetes, or pre-diabetes. The received seed values for the calibration parameters may be values based on one or more user-entered blood glucose levels. Displaying values corresponding to the received signals may include displaying a graph or table showing currently measured and historical values of the analyte concentration, detecting that a calibration change has occurred; , adjusting one or more calibration parameters of an analyte concentration sensor in accordance with a change in calibration and, following adjustment, presenting currently measured and historical values of the analyte concentration in accordance with the adjusted calibration parameters. and updating the display of the graph or table. Detecting that a calibration change has occurred may include detecting a slow moving average change or detecting a steady state value change.

In a sixth aspect, a method is provided for calibrating an analyte concentration sensor, wherein following sensor insertion into a patient only parameters based on or derivable from the sensor signal are used, at least an initial value of analyte concentration and receiving an initial value or initial value distribution of sensor sensitivity; monitoring the signal from the sensor for a duration following insertion of the analyte concentration sensor; calculating a plurality of analyte concentration values based on the initial value or value distribution of sensitivity; determining a value distribution of the monitored signal over time; optimizing the analyte concentration values to match the value distribution of the monitored signal; and determining an updated sensitivity based on the optimization.

Implementations of aspects and embodiments can include one or more of the following. The receiving may be receiving an initial distribution of sensor sensitivities, and the calculation of the plurality of analyte concentration values may be based on the monitored sensor signal and representative values from the initial distribution of sensor sensitivities. A representative value may be selected from the mean or midpoint or median. The updated sensitivity determination may further comprise dividing the representative value by the initial value of the analyte concentration and updating the sensitivity value to equal the result of the division. The initial value of the analyte population may be the population mean, entered by the user, or transferred from a previous session. The optimization may comprise optimizing the initial value or value distribution of sensor sensitivity and the product of multiple analyte concentration values. Product optimization optimizes the product to match the value distribution of the monitored signal while optimizing the parameters of the value distribution of the sensor sensitivity and the multiple analyte concentration values to the corresponding population mean and most stringent values. may include adjusting to match . Receiving may further comprise receiving a baseline initial value distribution, and optimizing the baseline value distribution together with the sensor sensitivity value distribution and the plurality of analyte concentration values to the values of the monitored signal. It may further comprise optimizing to match the distribution. The baseline initial value distribution may follow a normal distribution. At least the initial value of the analyte concentration may be used as part of the seed value input to the slow moving average filter. The initial value distribution of sensor sensitivity may be defined by a normal distribution. The determined value distribution of the monitored signal may follow a lognormal distribution. The duration may be one day. The method may further comprise continuing to determine the updated sensitivity based on the previous updated sensitivity and the received analyte concentration value. The method may further comprise detecting a slow moving average of the monitored analyte concentration values. If the absolute value of change in the slow moving average is greater than a predetermined threshold over a predetermined unit time, the method may include prompting the user to enter the data. If the absolute value of change in the slow moving average is greater than a predetermined threshold over a predetermined unit time, the method includes determining whether the change is due to a system error or a change in the actual sensitivity of the sensor. It's okay. Determining whether a change is due to a system error or a change in the actual sensitivity of the sensor involves determining whether the subsequent behavior of sensitivity is consistent with a known sensitivity profile, including the envelope of the sensitivity curve. may contain. If it is determined that the absolute value of the change in the slow moving average is due to the change in the actual sensitivity of the sensor, the method includes updating the sensitivity based at least in part on the value of the change in the slow moving average. It's okay. Determining whether the change is due to system error or due to changes in the actual sensitivity of the sensor determines whether the subsequent behavior of the analyte concentration values is consistent with the known envelope of physiological feasibility. may include If the absolute value of change in the slow moving average is determined to be due to system error, the method may include prompting the user to enter the data.

In a seventh aspect, a method of calibrating an analyte concentration sensor in a biological system using a signal from an analyte concentration sensor is provided to receive or determine seed values for calibration parameters associated with the analyte concentration sensor. using the seed value to at least partially determine calibration of the analyte concentration sensor; using the analyte concentration sensor to measure the analyte concentration after; and displaying the calibrated measurement using the seed value.

Implementations of aspects and embodiments can include one or more of the following. The receiving or determining may be performed on a monitoring device in operable signal communication with the analyte concentration sensor. The display may be performed on the monitoring device or on a mobile device in signal communication with the monitoring device. Displaying the measurements may include displaying a graph or table showing at least historical values of the analyte concentration, detecting that a calibration change has occurred, and performing an analysis according to the detected change in calibration. further adjusting one or more calibration parameters of the analyte concentration sensor; following the adjustment, updating the graphical or tabular display showing at least historical values of the analyte concentration according to the adjusted calibration parameters. may contain. An update may change the display of the historical value of the analyte concentration. The seed value may be based at least in part on the code. The code may be entered into the monitoring device by the user. The monitoring device may be configured to receive the code without substantial user involvement. The seed value may be based at least in part on impedance measurements. The seed value may be based at least in part on information associated with the manufacturing lot of the sensor. The seed value may be based at least in part on a population average. The seed value may be based at least in part on the user's most recent analyte value.

In an eighth aspect, a method is provided for calibrating and compensating for drift of an indwelling analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, wherein at steady state, the analyte in the biological system The concentration value is known, and the method comprises, on the monitoring device, detecting when the analyte concentration value measured by an analyte concentration sensor indwelling within the biological system is at steady state; or on a device or server operably linked to a monitoring device, correlating the measured analyte concentration values when the biological system is in a detected steady state with said known analyte concentration values; Determining a first slow moving average of analyte measurements over a first time period and basing calibration of the sensor based at least in part on the first slow moving average and known analyte concentration values. and, following the first determination, determining a second slow moving average of the analyte measurements over a second time period; and adjusting the calibration of the sensor based on the difference between .

In a ninth aspect, a method is provided for calibrating a first portion of a number of sensors, the second portions being in service and obtaining calibration data from some of the second portions of the sensors. receiving and updating one or more calibration parameters of the first portion based on the received data.

Implementations of aspects and embodiments can include one or more of the following. The update may be performed before the first portion is installed to the user. Updating may be performed after the first portion has been incorporated into the user. The update may be performed by sending new or updated calibration parameters over the network to the sensor electronics module associated with the monitoring device or sensor. A second portion of the sensor may be configured to be calibrated using an a priori calibration. A second portion of the sensor may be configured to be calibrated using user data. A second portion of the sensor may be configured to be calibrated using an ex vivo bench calibration. A second portion of the sensor may be configured to be calibrated using blood measurements.

In a tenth aspect, a method is provided for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, wherein the indwelling analyte concentration sensor is used to measuring the value as a function of time; filtering the measurement using a double exponential smoothing filter; following filtering, displaying the filtered measurement against time; including.

Implementations of aspects and embodiments can include one or more of the following. A double exponential smoothing filter may be governed by the equations described herein. A subsequent glucose signal as a function of time may be provided by the equations described herein.

In an eleventh aspect, there is provided a method of calibrating an analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, comprising: are known, and the method detects when, on the monitoring device, the set of analyte concentration values measured by an analyte concentration sensor indwelling within the biological system constitutes a repeatable event. and correlating, on the monitoring device or on a device or server operably coupled to the monitoring device, the set of analyte concentration values at the time of the repetitive event with known analyte concentration values; including.

Implementations may include being selected from the group consisting of steady state, postprandial rise, daily high and low glucose range, decay rate, or rate of change.

In a twelfth aspect, a method is provided for compensating for drift in an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, wherein the indwelling analyte concentration sensor is used to measuring a value and determining a first slow moving average of the analyte measurements over a first set of time periods, the first set including the event-based time period; and referencing the calibration of the sensor based at least in part on a first slow moving average; and following the first determination, a second slow moving analyte measurement over a second set of time periods. determining an average, wherein the second set includes an event-based period; determining, at least in part, a difference between the first slow moving average and the second slow moving average; and adjusting the calibration of the sensor based on.

Implementations can include one or more of the following. The first and second event-based time periods may be selected from the group consisting of a postprandial period, a sleep period, and a postbreakfast period.

In a thirteenth aspect, a method of calibrating an analyte concentration sensor in a biological system is provided, comprising determining a sensitivity profile over time for a set of sensors of a type; measuring an electrical characteristic of the transmitter; reading the identifier of the sensor; receiving data corresponding to the sensitivity of the sensor; and storing the data obtained.

Implementations can include one or more of the following. A set of sensors of a type may correspond to a set of sensors within a lot. The method may further include packaging the individual sensors with the transmitter as a kit. The method may further include coupling the transmitter to a mobile device running a surveillance application. The method may further include calibrating the transmitter and sensor using the monitoring application. Calibration may relate to measured electrical properties of the transmitter. The monitoring application may be configured to initiate the sensor session by a signal from the transmitter, the signal detecting that the transmitter is coupled to the sensor. The transmitter may be configured to initiate a sensor session when the transmitter detects a connection to the sensor. The method may further include coupling the transmitter to a mobile device running a surveillance application. The method may further comprise receiving a representative set of measured analyte values. The method may further comprise using the received representative set of measured analyte values or a subset thereof to determine seed parameters for the forward filter, the inverse filter, or both. The seed value may be determined using the signal median value, the drift value, or both. Both forward and backward filters may be used, and the method may further include optimizing the seed value to minimize the mean squared error between the two signal filters. The method may further include adjusting the sensitivity and baseline for the sensor according to a signal-based calibration algorithm, which averages the signals from the forward and reverse filters along with the raw sensor signal. to use. The method may further comprise adjusting the sensitivity and baseline based on one or more criteria. Criteria may include that the mean glucose value should match the predicted diabetes mean. Criteria may include that CGM glucose variability should be consistent with mean glucose levels. The method may further include detecting an amount of sensor change, determining that the amount of sensor change exceeds a threshold criterion, and preventing display of the reading, thereby potentially Inaccurate readings are not displayed to the user.

In a fourteenth aspect, a method is provided for compensating for analyte concentration sensor drift in a biological system using only the signal from the analyte concentration sensor, wherein the indwelling analyte concentration sensor is used to measuring a value; determining a first slow moving average of the analyte measurements over a first time period; and referencing calibration of the sensor based at least in part on the first slow moving average. and, following the first determination, determining a second slow moving average of the analyte measurements over a second period of time; determining at least partially the seed value and the first slow moving average and the second adjusting sensor calibration based on the difference between the slow moving average.

Implementations can include one or more of the following. The seed value may be determined using the signal median value, the drift value, or both. Both forward and backward filters may be used, and the method may further include optimizing the seed value to minimize the mean squared error between the two signal filters. The method may further include adjusting the sensitivity and baseline for the sensor according to a signal-based calibration algorithm, which averages the signals from the forward and reverse filters along with the raw sensor signal. to use. The method may further comprise adjusting the sensitivity and baseline based on one or more criteria. Criteria may include that the mean glucose value should match the predicted diabetes mean. Criteria may include that CGM glucose variability should be consistent with mean glucose levels.

In further aspects and embodiments, the above method features of various aspects are described in terms of systems as in various aspects configured to perform the method features. Any of the features of an embodiment of any of the above-referenced aspects, including, but not limited to, any embodiment of any of the above-referenced first to fourteenth aspects, can be It is applicable to all other aspects and embodiments identified herein, including, but not limited to, any embodiment of any of aspects 1-14. Further, any of the features of the embodiments of the various aspects, including but not limited to any embodiment of any of the first through fourteenth aspects referred to above, may be incorporated herein in any manner. Other embodiments described herein may be independently combined in part or in whole, e.g., one, two, three or more embodiments may be combined in whole or in part. could be. Furthermore, any of the features of the embodiments of the various aspects, including but not limited to any embodiment of any of the above mentioned aspects 1-14, may be applied to any other aspect or embodiment. can be made optional. Any aspect or embodiment of the method may be performed by a system or apparatus of another aspect or embodiment, and any aspect or embodiment of the system or apparatus may include any of the first to fourteenth aspects referred to above. Any such embodiment may be configured to perform the method of another aspect or embodiment, including but not limited to.

This Summary of the Invention is provided to introduce a selection of concepts in a simplified form. Concepts are further described in the detailed description. Elements or steps other than those described in this Summary of the Invention are possible, and no element or step is necessarily required. This Summary of the Invention is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to determine the scope of the claimed subject matter. Nor is it intended for use as an adjunct to The claimed subject matter is not limited to implementations that solve any or all disadvantages noted in any part of this disclosure.

This embodiment will now be considered in detail, with an emphasis on clarifying advantageous features. These embodiments illustrate novel and non-obvious sensor signal processing and calibration systems and methods illustrated in the accompanying drawings, which are for illustrative purposes only and are not to scale, rather emphasizing the principles of the disclosure. do. These drawings include the following figures, like numerals indicating like parts: FIG.

1 is a schematic diagram of a continuous analyte sensor system attached to a host and in communication with multiple exemplary devices; FIG. 2 is a block diagram showing the electronics associated with the sensor system of FIG. 1; FIG. FIG. 4 shows a graph showing a linear relationship between measured sensor counts and analyte concentration. indicates an exemplary change in sensitivity over time. 4A and 4B illustrate various methods of providing factory information about the sensor to the transmitter electronics. Fig. 10 shows a "no code" option for providing factory information about the sensor to the sensor electronics; Fig. 3 shows options for providing factory information about the sensor to the sensor electronics without explicit user input. 4 illustrates options for using user input to provide factory information about the sensor to the sensor electronics. FIG. 4 illustrates the steps of using one portion of many sensors, which is acquiring field data, to calibrate another portion of many sensors. FIG. 4 illustrates the steps of using one portion of many sensors, which is acquiring field data, to calibrate another portion of many sensors. FIG. 11 illustrates an exemplary method in accordance with the present principles and, in particular, is a flow chart for performing the method in accordance with FIGS. 9 and 10; FIG. FIG. 2 is a schematic diagram of sensors and transmitters in a host communicating with a receiver and/or smart phone; 4 is a flow chart illustrating another exemplary method in accordance with the present principles; FIG. 4 is a graph showing analyte concentration over time before sensitivity change. FIG. FIG. 10 is a graph showing analyte concentration over time after sensitivity changes; FIG. 1 is a modular diagram of an analyte concentration measurement system in accordance with present principles; FIG. 4 is a flow chart illustrating another exemplary method in accordance with the present principles; 5 is a flow chart illustrating another exemplary method consistent with the present principles for performing calibration using steady state; Fig. 10 is a graph showing two calibration lines before and after drift; A slow moving average of sensor counts over time is shown. A slow moving average of sensor counts over time is shown. 5 is a flow chart illustrating another exemplary method in accordance with present principles using a slow moving average; 5 is a flowchart illustrating another exemplary method consistent with the present principles showing updating of history values; 5 is a chart showing sensitivity data over an extended sensor session showing characteristic drift. 5 is a chart showing sensitivity data over an extended sensor session showing characteristic drift with failure mode. 4 is a flow chart illustrating another exemplary method in accordance with the present principles; 4 is a flow chart illustrating another exemplary method in accordance with the present principles; 4 is a flow chart illustrating another exemplary method in accordance with the present principles; 4 is a flow chart illustrating another exemplary method in accordance with the present principles; 4 is a graph showing the distribution of sensitivity; is a graph showing the baseline distribution. 1 is a graph showing a distribution of glucose levels, eg, glucose levels over time. FIG. 10 is a graph showing the most likely sensitivity (slope) values given exemplary parameters; FIG. FIG. 4 is a graph showing the most likely baseline values given exemplary parameters; FIG. FIG. 4 is a graph showing the most likely glucose values given exemplary parameters; FIG. 1 shows an exemplary glucose trace. 1 shows an exemplary glucose trace. We also show the effect of a double exponential filter acting on the glucose trace. Figure 36 shows an estimated drift curve for the sensor used to determine Figures 34 and 35; Fig. 4 is a further chart showing drift correction according to the above principles; Fig. 4 is a further chart showing drift correction according to the above principles; Fig. 4 is a further chart showing drift correction according to the above principles; Fig. 4 is a further chart showing drift correction according to the above principles; Fig. 2 shows the measured relationship between signal coefficient of variation and standard deviation of glucose concentration values; 4 shows a measured glucose concentration signal over a period of time. Fig. 2 shows a linear relationship between signal coefficient of variation and glucose concentration value standard deviation, with exemplary values marked; Shows distribution of differences between measured and predicted standard deviations. Figure 2 shows the relationship between mean glucose and glucose standard deviation. Differences between glucose standard deviations are shown between patient populations, ie, non-diabetic, type I diabetic, and type II diabetic. Data points separated by a time lag are indicated, with Δ representing the individual rate of change between two adjacent points. 4 is a flow chart illustrating another implementation of a method in accordance with the present principles;

The following description and examples further illustrate certain exemplary embodiments of the disclosed invention. Those skilled in the art will recognize that there are many variations and modifications of this invention that are encompassed by its scope. Therefore, the description of certain exemplary embodiments should not be construed as limiting the scope of the invention.

Definitions To facilitate understanding of the preferred embodiments, a number of terms are defined below.

As used herein, the term "analyte" is a broad term, the ordinary and customary meaning of which is given to those of ordinary skill in the art (and the special or customized meaning). ), and also refers to, but is not limited to, substances or chemical constituents in bodily fluids that may be analyzed, such as blood, interstitial fluid, cerebrospinal fluid, lymph, or urine. Analytes can include naturally occurring substances, man-made substances, metabolites, and/or reaction products. In some embodiments, the analyte for measurement by the sensor heads, devices and methods is glucose. However, other analytes are contemplated as well, acarboxyprothrombin, acylcarnitine, adenine phosphoribosyltransferase, adenosine deaminase, albumin, α-fetoprotein, amino acid profiles (arginine (Krebs cycle), histidine/urocanic acid, homocysteine, phenylalanine/tyrosine, tryptophan), andrenostenedione, antipyrine, arabinitol enantiomers, arginase, benzoylecgonine (cocaine), biotinidase, biopterin, c-reactive protein, carnitine, carnosinase, CD4, ceruloplasmin, chenodeoxycholic acid, chloroquine , cholesterol, cholinesterase, conjugated 1-β-hydroxycholic acid, cortisol, creatine kinase, creatine kinase MM isozyme, cyclosporin A, d-penicillamine, de-ethylchloroquine, dehydroepiandrosterone sulfate, DNA (acetylation polymorphism, alcohol dehydrogenase, α1-antitrypsin, cystic fibrosis, Duchenne/Becker muscular dystrophy, analyte-6-phosphate dehydrogenase, hemoglobin A, hemoglobin S, hemoglobin C, hemoglobin D, hemoglobin E, hemoglobin F, D Punjab, β-thalassemia, hepatitis B virus, HCMV, HIV-1, HTLV-1, Leber hereditary optic neuropathy, MCAD, RNA, PKU, Plasmodium vivax, sexual differentiation, 21-deoxycortisol), desbutyl Halofantrin, dihydropteridine reductase, diphtheria/tetanus antitoxin, erythrocyte arginase, erythrocyte protoporphyrin, esterase D, fatty acid/acylglycine, free β-human chorionic gonadotropin, free erythrocyte porphyrin, free thyroxine (FT4), free tri- Iodothyronine (FT3), fumarylacetoacetase, galactose/gal-1-phosphate, galactose-1-phosphate uridyltransferase, gentamicin, analyte-6-phosphate dehydrogenase, glutathione, glutathione peroxidase, glycochol acid, glycosylated hemoglobin, halofantrin, hemoglobin variant, hexosaminidase A, human erythrocyte carbonic anhydrase I, 17-α-hydroxyprogesterone, hypoxanthine phosphoribosyltransferase, immunoreactive trypsin, milk acid, lead, lipoprotein ((a), B/A-1, β), lysozyme, mefloquine, netilmicin, phenobarbitone, phenytoin, phytanic acid/pristanic acid, progesterone, prolactin, prolidase, purine nucleoside phosphorylase, quinine , inverted tri-iodothyronine (rT3), selenium, serum pancreatic lipase, sisomycin, somatomedin C, specific antibodies (adenovirus, anti-nuclear antibody, anti-zeta antibody, arbovirus, Aujesky virus, dengue virus, guinea worm, Taenia uniculatus, Entamoeba histolytica, Enterovirus, Giardia duodenalisa, Helicobacter pylori, Hepatitis B virus, Herpes virus, HIV-1, IgE (atopic disease), Influenza virus, Leishmania donovan, Leptospira, Measles / Mumps / Rubella, Mycobacterium leprae, Mycoplasma pneumoniae, Myoglobin, Parainfluenza virus, Plasmodium falciparum, Poliovirus, Pseudomonas aeruginosa, Respiratory rash virus, Rickettsia (tsutsugamushi disease), Manson Schistosoma, Toxoplasma gondii, Trepenoma pallidium, Trypanosoma cruzi/Langer, Vesicular stomatis virus, Bancroft's heartworm, Yellow fever virus), specific antigens (hepatitis B virus, HIV- 1), succinylacetone, sulfadoxine, theophylline, thyrotropine (TSH), thyroxine (T4), thyroxine-binding globulin, trace elements, transferrin, UDP-galactose-4-epimerase, urea, uroporphyrinogen I synthase, vitamin A , leukocytes, and zinc protoporphyrin. Salts, sugars, proteins, fats, vitamins, and hormones naturally occurring in blood or interstitial fluid may also constitute analytes in certain embodiments. Analytes, such as metabolites, hormones, antigens, antibodies, etc., can be naturally present in bodily fluids. Alternatively, analytes can be introduced into the body, such as contrast agents for diagnostic imaging, radioisotopes, chemical agents, fluorocarbon-based artificial blood, or drugs or pharmaceutical compositions, such as insulin, ethanol, cannabis (marijuana , tetrahydrocannabinol, hashish), inhalants (nitrous oxide, amyl nitrite, butyl nitrite, chlorohydrocarbons, hydrocarbons), cocaine (crack cocaine), stimulants (amphetamine, methamphetamine, Ritalin, Cylert, Preludin, Didrex, PreState, Voranil, Sandrex, Plegine), depressants (barbiturates, mechalons, tranquilizers such as Valium, Librium, Miltown, Serax, Equanil, Tranxene), hallucinogens (phencyclidine, lysergic acid, mescaline, peyote, psilocybin), narcotics (heroin, codeine, morphine, opium, meperidine, Percocet, Percodan, Tussionex, Fentanyl, Darvon, Talwin, Lomotil), designer drugs (fentanyl, meperidine, amphetamine, methamphetamine, and phencyclidine). analogues such as Ecstasy), anabolic steroids, and nicotine. Metabolites of drugs and pharmaceutical compositions may also be contemplated as analytes. For example, ascorbic acid, uric acid, dopamine, noradrenaline, 3-methoxytyramine (3MT), 3,4-dihydroxyphenylacetic acid (DOPAC), homovanillic acid (HVA), 5-hydroxytryptamine (5HT), and 5-hydroxyindoleacetic acid Analytes such as neurochemicals and other chemicals produced in the body, such as (FHIAA), can be analyzed.

As used herein, the terms "microprocessor" and "processor" are broad terms, the ordinary and customary meanings of which are given to those skilled in the art (and the specific or in a customized sense), and furthermore, a computer system that performs arithmetic and logic operations using logic circuits that respond to and process primitive instructions that drive the computer, state machines etc., but not limited to these.

As used herein, the terms "raw data stream" and "data stream" are broad terms and their ordinary and customary meanings are given to those skilled in the art ( and without being limited to any special or customized meaning) and further refers to, but is not limited to, analog or digital signals directly related to glucose measured from a glucose sensor. In one example, the raw data stream is digital data in "counts" (e.g., voltage or amperage) converted by an A/D converter from an analog signal to provide one or more data points representing glucose concentration. include. These terms broadly encompass multiple time-spaced data points from a substantially continuous glucose sensor, with time intervals ranging from less than 1 second up to, for example, 1, 2, or 5 minutes or more. Includes individual measurements taken. In another example, the raw data stream includes integrated digital values, where the data includes one or more data points representing a glucose sensor signal averaged over a period of time.

As used herein, the term "proofreading" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning of which is without limitation), and further, in real time, with or without reference data, a process of determining relationships between sensor data and corresponding reference data, which Refers to, but is not limited to, processes that can be used to transform reference data into meaningful values that are substantially equivalent. In some embodiments, i.e., in continuous analyte sensors, calibration is performed over time as changes in, for example, sensitivity, baseline, transport, metabolism, etc., result in changes in the relationship between sensor data and reference data. can be updated or recalibrated dynamically (at the factory, in real time, and/or retroactively). Calibration can also be accomplished by automatically predicting sensor signal parameters through analysis of one or more signal characteristics or features (autocalibration).

As used herein, the terms "calibrated data" and "calibrated data stream" are broad terms and their ordinary and customary meanings are given to those skilled in the art. Yes (and not limited to any special or customized meaning), and using functions, including through the use of sensitivity, e.g., transform functions, to provide meaningful value to the user , refers to, but is not limited to, data that has been transformed from its raw state to another state.

As used herein, the terms "smoothed data" and "filtered data" are broad terms and their ordinary and customary meanings are given to those skilled in the art. Yes (and not limited to any special or customized meaning), and also to make the data smoother and more continuous by performing moving averages on the raw data stream, including, for example, slow moving averages It refers to, but is not limited to, data that has been modified to make it more general and/or remove or reduce anomalies. Examples of data filters include, but are not limited to, FIR (finite impulse response), IIR (infinite impulse response), moving average filters, and the like.

As used herein, the terms "smoothing" and "filtering" are broad terms, the ordinary and customary meanings of which are given to those skilled in the art (and the special or customized semantics), and also to make the data set smoother and more continuous, for example by performing a moving average of the raw data stream, or anomalous points Refers to, but is not limited to, modifying a data set to remove or reduce .

As used herein, the term "algorithm" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning of without limitation), and also refers to, but is not limited to, computational processes (e.g., programs) involved in transforming information from one state to another, for example, by using computational processes not.

As used herein, the term "count" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning of without limitation) and furthermore, but without limitation, a unit of measurement for a digital signal. In one example, the raw data stream measured in counts is directly related to voltage (eg, converted by an A/D converter), which is directly related to current from the working electrode.

As used herein, the term "sensor" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning of and without limitation) and further refers to, but is not limited to, a component or region of the device in which an analyte can be quantified.

As used herein, the terms "glucose sensor" and "member for determining the amount of glucose in a biological sample" are broad terms and are subject to their ordinary and customary meanings. (and not limited to any particular or customized meaning) and further refers to any mechanism (e.g., enzymatic or non-enzymatic) by which glucose can be quantified. , but not limited to. For example, some embodiments utilize membranes containing glucose oxidase that catalyze the conversion of oxygen and glucose to hydrogen peroxide and gluconate as exemplified by the following chemical reactions.

glucose _{+ O2} → gluconate ₊ H2O2

For _each glucose molecule metabolized, there is a proportional change in co _- reactant _O2 and product H2O2, so electrodes are used to monitor current changes in either co-reactant or product to determine the glucose concentration.

As used herein, the terms "operably connected" and "operably linked" are broad terms and their ordinary and customary meanings are given to those skilled in the art. (and not limited to any special or customized meaning), and is coupled to another component(s) in a manner that permits the transmission of signals between the components Refers to, but is not limited to, one or more components. For example, one or more electrodes can be used to detect the amount of glucose in a sample and convert that information into a signal, such as an electrical or electromagnetic signal, which can then be transmitted to an electronic circuit. . In this case, the electrodes are "operably linked" to the electronic circuit. These terms are broad enough to include wireless connections.

The term "determining" encompasses a wide range of actions. For example, "determining" means calculating, operating, processing, deriving, examining, searching (e.g., searching in a table, database, or another data structure); It can include determining, and the like. Also, "determining" can include receiving (eg, receiving information), accessing (eg, accessing data in memory), and the like. Also, "determining" may include resolving, selecting, selecting, calculating, deriving, establishing, and/or the like. Determining may also include establishing that the parameters meet predetermined criteria, including meeting, passing, exceeding, etc. thresholds.

As used herein, the term "substantially" is a broad term, the ordinary and customary meaning of which is given to those of ordinary skill in the art (and the special or customized not limited in meaning), and also refers primarily to, but not necessarily fully specified by, but not limited to.

As used herein, the term "host" is a broad term, the ordinary and customary meaning of which is given to those of skill in the art (and the special or customized meaning of which is without limitation), and also refers to mammals, particularly but not limited to humans.

As used herein, the term "continuous analyte (or glucose) sensor" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special , or in a customized sense), and further, the concentration of the analyte continuously or continuously at time intervals ranging from, e.g., less than 1 second up to, e.g., 1, 2, or 5 minutes or more. It refers to, but is not limited to, a device that measures objectively. In one exemplary embodiment, the continuous analyte sensor is a glucose sensor as described in US Pat. No. 6,001,067, which is incorporated herein by reference in its entirety.

As used herein, the term "continuous analyte (or glucose) sensing" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the specific , or in a customized sense), and furthermore, the monitoring of the analyte is continuous or It refers to a period of continuous implementation, but is not limited to this.

As used herein, the terms “reference analyte monitor,” “reference analyte meter,” and “reference analyte sensor” are broad terms and their ordinary and customary meanings are Devices that are shown to those skilled in the art (and are not limited in any special or customized sense) and that also measure the concentration of an analyte and can be used as a reference for continuous analyte sensors. For example, but not limited to, a self-monitoring blood glucose meter (SMBG) can be used as a reference for a continuous glucose sensor for comparison, calibration, and the like.

As used herein, the term "sensing membrane" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning). and may consist of two or more regions, typically permeable to oxygen and may or may not be permeable to glucose, a number Refers to, but is not limited to, permeable or semi-permeable membranes composed of materials that are microns or thicker. In one example, the sensing membrane contains immobilized glucose oxidase enzyme, which allows an electrochemical reaction to occur to measure the concentration of glucose.

As used herein, the term "physiologically possible" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the specific, or customized meaning) and further refers to, but is not limited to, physiological parameters obtained from continuous studies of glucose data in humans and/or animals. For example, the maximal sustained rate of change of glucose in humans of about 4-5 mg/dL/min and the maximal acceleration of the rate of change of about 0.1-0.2 mg/dL/min/min is at the physiologically possible limit. is considered. Values outside these limits, for example, are considered non-physiological and are likely the result of signal errors. As another example, the rate of change of glucose is lowest at the maximum and minimum of the daily glucose range, which range is the region of greatest risk in patient treatment, thus the physiologically possible rate of change is , can be set at maximum and minimum based on continuous study of glucose data. As a further example, it has been observed that the best solution for the shape of the curve at any point along the glucose signal data stream over a certain period of time (eg, about 20-30 minutes) is a straight line, which is It can be used to set physiological limits.

As used herein, the term "frequency component" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning). , and also refers to spectral density, including but not limited to frequencies contained within signals and their power.

As used herein, the term "linear regression" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning). ), and further refers to, but is not limited to, finding the line whose data set has the smallest measured deviation or separation from that line. The by-products of this algorithm include slope, y-intercept, and R-squared values that determine how well the measured data fits the line. In certain cases, robust regression techniques can also be used to handle outliers in regression.

As used herein, the term "nonlinear regression" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized means, but is not limited to, fitting a set of data to describe the relationship between a response variable and one or more explanatory variables in a non-linear manner. .

As used herein, the term "average" is a broad term, the ordinary and customary meaning of which is given to those of ordinary skill in the art (and the special or customized meaning of without limitation) and further refers to, but is not limited to, dividing the sum of observations by the number of observations.

As used herein, the term "non-recursive filter" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized ) and further refers to, but is not limited to, equations that use moving averages as inputs and outputs.

As used herein, the terms "recursive filter" and "autoregressive algorithm" are broad terms whose ordinary and customary meanings are given to those skilled in the art (and without being limited to any special or customized meaning) and also refers to, but is not limited to, equations where the previous average is part of the next filtered output. More specifically, the generation of a sequence of observations in which the value of each observation depends in part on the value of its immediately preceding observation. One example is a regression structure in which delayed response values play the role of independent variables.

As used herein, the term "variation" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning of without limitation) and further refers to, but is not limited to, the amount of difference or variation from a point, line, or set of data. In one embodiment, the estimated analyte value can have variability including, for example, a range of values other than the estimated analyte value representing a range of possibilities based on known physiological patterns.

As used herein, the terms "physiological parameter" and "physiological boundary" are broad terms and their ordinary and customary meanings are given to those skilled in the art (and without being limited to any special or customized meaning) and furthermore refers to, but is not limited to, parameters obtained from continuous studies of physiological data in humans and/or animals. For example, a maximum sustained rate of change of glucose in humans of about 6-8 mg/dL/min and a maximum acceleration rate of change of about 0.1-0.2 mg/dL/min2 ^are physiologically possible limits. values outside these limits are considered non-physiological. As another example, the rate of change of glucose is lowest at the maximum and minimum of the daily glucose range, which range is the region of greatest risk in patient treatment, thus the physiologically possible rate of change is , can be set at maximum and minimum based on continuous study of glucose data. As a further example, it has been observed that the best solution for the shape of the curve at any point along the glucose signal data stream over a certain period of time (eg, about 20-30 minutes) is a straight line, which is It can be used to set physiological limits. These terms are broad enough to include the physiological parameters associated with any analyte.

As used herein, the term "measured analyte value" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the specific or and without limitation to a customized meaning) and further refers to, but is not limited to, an analyte value or set of analyte values for the time period over which the analyte data was measured by the analyte sensor. The term is broad enough to include data from an analyte sensor before or after data processing (eg, data smoothing, calibration, etc.) in the sensor and/or receiver.

As used herein, the term "estimated analyte value" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized ) and further refers, without limitation, to an analyte value or set of analyte values that is algorithmically extrapolated from measured analyte values.

As used herein, the term "sensor data" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized meaning). ) and further refers to any data associated with a sensor, such as, but not limited to, a continuous analyte sensor. Sensor data is the raw data stream of analog or digital signals directly related to the measured analyte from the analyte sensor (or other signals received from another sensor), or simply the data stream, as well as the calibration and/or filtered raw data. In one example, the sensor data includes digital data in "counts" (eg, voltage or amperage) converted from an analog signal by an A/D converter and includes one or more data points representing glucose concentration. As such, the terms "sensor data point" and "data point" generally refer to a digital representation of sensor data at a particular time. These terms broadly encompass multiple time-spaced data points from a sensor, such as from a substantially continuous glucose sensor, ranging from less than 1 second up to, for example, 1, 2, or 5 minutes or more. It contains individual measurements taken over a range of time intervals. In another example, sensor data includes integrated digital values representing one or more data points averaged over a period of time. Sensor data may include calibrated data, smoothed data, filtered data, transformed data, and/or any other data associated with the sensor.

As used herein, the terms "matched data pair" or "data pair" are broad terms, the ordinary and customary meaning of which is given to those of ordinary skill in the art (and the special specific or customized meaning), and further reference data (e.g., one or more Reference analyte data points), but not limited to.

As used herein, the term "sensor electronics" is a broad term, the ordinary and customary meaning of which is given to those skilled in the art (and the special or customized without limitation in meaning), refers to but is not limited to a component (eg, hardware and/or software) of a device configured to process data.

As used herein, the term "calibration set" is a broad term, the ordinary and customary meaning of which is given to those of ordinary skill in the art (and the special or customized meaning). ), refers to, but is not limited to, a set of data that contains information useful for calibration. In some embodiments, the calibration set is formed from one or more matched data pairs used to determine the relationship between reference data and sensor data, but prior to embedding, externally , or other internally obtained data may also be used. As another example, data may also be used from previous sensor sessions of the target user.

As used herein, the terms "sensitivity" or "sensor sensitivity" are broad terms, the ordinary and customary meanings of which are given to those skilled in the art (and the specific or a customized meaning), the measured analyte, or the measured species ₍ e.g., H2O2) associated with the measured analyte ₍ e.g., glucose). It refers to, but is not limited to, the amount of signal produced by concentration. For example, in one embodiment, the sensor has a sensitivity of about 1 to about 300 picoamperes of current per mg/dL of glucose analyte.

As used herein, the terms "sensitivity profile" or "sensitivity curve" are broad terms, the ordinary and customary meanings of which are given to those skilled in the art (and the special , or in a customized sense), refers to, but is not limited to, an indication of changes in sensitivity over time.

Other definitions are provided within the description below, sometimes out of the context of the term's use.

As used herein, the following abbreviations apply. Eq and Eqs (equivalents), mEq (milliequivalents), M (mole), mM (millimoles), μM (micromoles), N (norm), mol (moles), mmol (millimoles), μmol (micromoles), nmol (nanomole), g (gram), mg (milligram), μg (microgram), Kg (kilogram), L (liter), mL (milliliter), dL (deciliter), μL (microliter), cm (centimeter meters), mm (millimeters), μm (micrometers), nm (nanometers), h and hr (hours), min (minutes), and sec. (seconds), °C (Celsius).

System Overview/Overview Traditional in vivo continuous analyte sensing technologies have typically relied on reference measurements performed during a sensor session for calibration of continuous analyte sensors. The reference measurements are substantially matched with time-corresponding sensor data to create matched data pairs. A regression is then performed on the matched data pairs (eg, by using least-squares regression) to generate a transfer function that defines the relationship between the sensor signal and the estimated glucose concentration.

In critical care situations, calibration of continuous analyte sensors is often performed by using a calibration solution with a known concentration of analyte as a reference. This calibration procedure can be cumbersome because a calibration bag separate (and in addition to) the IV (intravenous) bag is typically used. In ambulatory settings, calibration of continuous analyte sensors is conventionally performed by peripheral blood glucose measurements (e.g., fingerstick glucose test), through which reference data is obtained and input to the continuous analyte sensor system. . This calibration procedure typically involves frequent fingerstick measurements, which can be inconvenient and painful.

To date, systems and methods for in vitro calibration (e.g., factory calibration) of continuous analyte sensors by the manufacturer that do not rely on periodic recalibration are mostly limited to the high level of sensor accuracy required for glycemic control. was inadequate for Part of this may be due to changes in sensor characteristics (eg, sensor sensitivity) that may occur during sensor use. Accordingly, calibration of continuous analyte sensors has typically involved periodic input of reference data, whether they are associated with calibration fluids or fingerstick measurements. This can be very cumbersome for users in everyday life and for patients in ambulatory situations or hospital staff in critical care situations.

The following description and examples describe the present embodiments with reference to the drawings. In the drawings, reference numbers indicate elements of this embodiment. These reference numbers are reproduced below in connection with the discussion of the corresponding drawing features.

Calibrate a continuous analyte sensor capable of achieving a high level of accuracy without (or with reduced reliance on) reference data from a reference analyte monitor (e.g., from a blood glucose meter) Systems and methods for are described herein.

Sensor System FIG. 1 illustrates an example system 100 according to some example implementations. System 100 includes a continuous analyte sensor system 8 that includes sensor electronics 12 and continuous analyte sensor 10 . System 100 may include other devices and/or sensors such as drug delivery pump 2 and glucose meter 4 . The continuous analyte sensor 10 may be physically connected to the sensor electronics 12 and may be integral (e.g., non-releasably attached) with the continuous analyte sensor 10 or may be releasably attached. It may be attachable. Sensor electronics 12 , drug delivery pump 2 and/or glucose meter 4 may be coupled with one or more devices such as display devices 14 , 16 , 18 and/or 20 .

In some exemplary implementations, system 100 communicates with a host (also referred to as a patient) from other devices, such as sensor system 8 and display devices 14-20, over network 406 (e.g., wired, wireless, or a combination thereof) to analyze the analyte data (and/or other patient-related data) and provide higher-level data such as statistical data about the analytes measured over a certain time frame may include a cloud-based analyte processor 490 configured to generate a report providing information on the A full discussion of using a cloud-based analyte processing system is provided by US patent entitled "Cloud-Based Processing of Analyte Data," filed March 7, 2013, which is incorporated herein by reference in its entirety. 13/788,375.

In some exemplary implementations, sensor electronics 12 may include electronic circuitry associated with measuring and processing data generated by continuous analyte sensor 10 . This generated continuous analyte sensor data may also include algorithms that may be used to process and calibrate the continuous analyte sensor data, although these algorithms may be provided in other ways as well. Sensor electronics 12 may include hardware, firmware, software, or a combination thereof to provide analyte level measurements via a continuous analyte sensor, such as a continuous glucose sensor. An exemplary implementation of sensor electronics 12 is further described below with respect to FIG.

Sensor electronics 12 may be coupled (eg, wirelessly, etc.) to one or more devices, such as display devices 14, 16, 18, and/or 20, as described. Display devices 14, 16, 18, and/or 20 present information such as sensor information transmitted by sensor electronics 12 for display on display devices 14, 16, 18, and/or 20 (and and/or issue an alert thereof).

The display device is a relatively small key fob display device 14 configured to present at least information (e.g., medication delivery information, separate self-monitoring glucose readings, heart rate monitors, food intake monitors, etc.). It may include a large handheld display device 16, a mobile phone 18 (eg, smart phone, tablet, etc.), computer 20, and/or any other user device.

In some exemplary implementations, the relatively small key fob-type display device 14 is used for wristwatches, belts, necklaces, pendants, jewelry, adhesive patches, pagers, key fobs, plastic cards (e.g., credit cards), identification ( ID) cards, and/or the like. The small display device 14 may include a relatively small display (e.g., smaller than the large display device 16) to display certain types of displayable sensor information such as numeric values, arrows, or color codes. It may be configured as

In some example implementations, relatively large handheld display devices 16 may include handheld receiver devices, palmtop computers, and/or the like. The large display device may include a relatively large display (e.g., larger than small display device 14) to display information such as graphical representations of continuous sensor data, including current and historical sensor data output by sensor system 8. may be configured to display.

In some exemplary implementations, continuous analyte sensor 10 includes a sensor for detecting and/or measuring an analyte, and continuous analyte sensor 10 includes non-invasive devices, subcutaneous devices, transdermal devices, and/or as an intravascular device, may be configured to continuously detect and/or measure an analyte. In some exemplary implementations, continuous analyte sensor 10 may analyze multiple intermittent blood samples, although other analytes may be used as well.

In some exemplary implementations, the continuous analyte sensor 10 uses one or more measurement techniques such as enzymatic, chemical, physical, electrochemical, spectrophotometric, polarized, calorimetric, iontophoretic, radiometric, immunochemical, and the like. may include a glucose sensor configured to measure glucose in blood or interstitial fluid. In implementations where the continuous analyte sensor 10 includes a glucose sensor, the glucose sensor includes any device capable of measuring the concentration of glucose, including invasive, minimally invasive, and non-invasive sensing techniques (e.g., fluorescence monitoring). ) may be used to provide data, such as a data stream indicative of the concentration of glucose in the host. The data stream may be sensor data (raw and/or filtered), which may be used by a user, patient, or caregiver (e.g., a parent, relative, guardian, It may be converted into a calibrated data stream that is used to provide glucose values to a host such as a teacher, doctor, nurse, or any other individual. Additionally, the continuous analyte sensor 10 may be implanted as at least one of the following types of sensors. Implantable glucose sensors, transcutaneous glucose sensors implanted in the host vessel or externally, subcutaneous sensors, replaceable subcutaneous sensors, intravascular sensors.

Although the disclosure herein refers to some implementations including continuous analyte sensors 10 including glucose sensors, continuous analyte sensors 10 may include other types of analyte sensors as well. Additionally, while some implementations refer to the glucose sensor as an implantable glucose sensor, other types of devices capable of detecting the concentration of glucose and providing an output signal representative of the glucose concentration as well. can be used. Furthermore, although the description herein refers to glucose as the analyte being measured, treated, etc., e.g. Other analytes can be used as well, including gon, acetyl-CoA, triglycerides, fatty acids, intermediates in the citric acid cycle, choline, insulin, cortisol, testosterone, and the like.

FIG. 2 illustrates an example of sensor electronics 12 according to some example implementations. Sensor electronics 12 may include sensor electronics configured to process sensor information, such as sensor data, and to generate, for example, converted sensor data and displayable sensor information via a processor module. . For example, the processor module may transform sensor data into one or more of the following: filtered sensor data (e.g., one or more filtered analyte concentration values), raw sensor data, calibrated sensor data (e.g., one or more calibrated analyte concentration values), rate of change information, trend information, acceleration/deceleration information, sensor diagnostic information, location information, alarm/warning information, calibration information, sensor data smoothing and/or filtering algorithms, and/or the like.

In some embodiments, processor module 214 is configured to accomplish a substantial portion, if not all, of the data processing. Processor module 214 may be integral to sensor electronics 12 and/or remotely located, such as one or more of devices 14 , 16 , 18 , and/or 20 and/or cloud 490 . good too. In some embodiments, processor module 214 may include multiple smaller subcomponents or submodules. For example, processor module 214 may include an alert module (not shown), or a prediction module (not shown), or any other suitable module that may be utilized to efficiently process data. If the processor module 214 is composed of multiple sub-modules, the sub-modules include within the sensor electronics 12 or other related devices (eg, 14, 16, 18, 20, and/or 490). may be located within For example, in some embodiments, processor module 214 may be located at least partially within cloud-based analyte processor 490 or elsewhere within network 406 .

In some example implementations, the processor module 214 may be configured to calibrate the sensor data, and the data storage device 220 may store the calibrated sensor data points as transformed sensor data. good. Further, processor module 214 wirelessly receives calibration information from display devices such as devices 14, 16, 18, and/or 20 to calibrate sensor data from sensor 12 in some exemplary implementations. may be configured to allow Additionally, processor module 214 may be configured to perform additional algorithmic processing on sensor data (e.g., calibrated and/or filtered data and/or other sensor information), and data storage 220 may: It may be configured to store the transformed sensor data and/or sensor diagnostic information associated with the algorithm.

In some example implementations, sensor electronics 12 may comprise an application specific integrated circuit (ASIC) 205 coupled to user interface 222 . ASIC 205 includes potentiostat 210, telemetry module 232 for transmitting data from sensor electronics 12 to one or more devices, such as devices 14, 16, 18, and/or 20, and/or signal processing and data storage. may further include other components for (eg, processor module 214 and data storage device 220). FIG. 2 shows an ASIC 205, a Field Programmable Gate Array (FPGA), one or more microprocessors configured to provide some (but not all) of the processing performed by the sensor electronics 12. Other types of circuits may be used as well, including analog circuits, digital circuits, or combinations thereof.

In the embodiment shown in FIG. 2, potentiostat 210 is coupled to continuous analyte sensor 10, such as a glucose sensor, to generate sensor data from the analyte. Potentiostat 210 also provides a voltage to continuous analyte sensor 10 via data line 212 to bias the sensor for measurement of a value (e.g., current value, etc.) indicative of analyte concentration within the host. (also referred to as the analog portion of the sensor). Potentiostat 210 may have one or more channels depending on the number of working electrodes in continuous analyte sensor 10 .

While in some example implementations potentiostat 210 may include a resistor that converts a current value from sensor 10 to a voltage value, in some example implementations a current-to-frequency converter (Fig. not shown) may also be configured to continuously incorporate measured current values from the sensor 10 using, for example, a charge counting device. In some exemplary implementations, an analog-to-digital converter (not shown) may digitize the analog signal from sensor 10 into so-called “counts” for processing by processor module 214 . The resulting counts may be directly related to the current measured by potentiostat 210, which may be directly related to analyte levels such as glucose levels within the host.

Telemetry module 232 may be operatively connected to processor module 214 and enables wireless communication between sensor electronics 12 and one or more other devices such as display devices, processors, network access devices, and the like. hardware, firmware, and/or software may be provided. Various wireless technologies that may be implemented in the telemetry module 232 include Bluetooth®, Bluetooth® Low-Energy, ANT, ANT+, ZigBee, IEEE 802.11, IEEE 802.16, cellular radio access technologies, Radio frequency (RF), infrared (IR), paging network communications, magnetic induction, satellite data communications, spread spectrum communications, frequency hopping communications, near field communications, and/or the like. In some exemplary implementations, telemetry module 232 includes a Bluetooth® chip, although Bluetooth® technology may also be implemented in the combination of telemetry module 232 and processor module 214 .

Processor module 214 may control the processing performed by sensor electronics 12 . For example, processor module 214 may process data (eg, counts) from sensors, filter data, calibrate data, perform fail-safe checks, and/or the like. may be configured to do things.

In some example implementations, the processor module 214 may comprise digital filters such as, for example, infinite impulse response (IIR) or finite impulse response (FIR) filters. This digital filter may smooth the raw data stream received from sensor 10 . Generally, digital filters are programmed to filter data sampled at predetermined time intervals (also called sampling rate). In some exemplary implementations, such as when potentiostat 210 is configured to measure an analyte (eg, glucose and/or the like) at discrete time intervals, these time intervals are determined by digital filtering. Determine the sampling rate. In some exemplary implementations, potentiostat 210 may be configured to continuously measure the analyte using, for example, a current-to-frequency converter. In these current-to-frequency converter implementations, the processor module 214 may be programmed to request digital values from the integrators of the current-to-frequency converter at predetermined time intervals (acquisition time). These digital values obtained by the processor module 214 from the integrators may be averaged over the acquisition time by the continuity of current measurements. Therefore, the acquisition time may be determined by the sampling rate of the digital filter. Other uses of FIR filters are described in more detail below.

Processor module 214 may further include a data generator (not shown) configured to generate data packages for transmission to devices such as display devices 14 , 16 , 18 and/or 20 . Additionally, processor module 214 may generate data packets for transmission to these external sources via telemetry module 232 . In some example implementations, the data package may be customizable for each display device as described and/or may include timestamps, displayable sensor information, converted sensor data, sensor and /or any use of identifier codes, raw data, filtered data, calibrated data, rate of change information, trend information, error detection or correction, and/or the like for sensor electronics 12; May contain possible data.

Processor module 214 may also include program memory 216 and other memory 218 . Processor module 214 may be coupled to a communication interface such as communication port 238 and a power source such as battery 234 . Additionally, the battery 234 may be further coupled to a charger and/or regulator 236 to provide power to the sensor electronics 12 and/or charge the battery 234 .

Program memory 216 is for storing data such as an identifier (e.g., sensor identifier (ID)) for coupled sensor 10 and for performing one or more of the operations/functions described herein. It may be implemented as a pseudo-static memory for storing code (also called program code) for calibrating the ASIC 205 to perform. For example, the program code may configure processor module 214 to process and filter data streams or counts, perform calibration methods described below, perform fail-safe checks, and so on.

Memory 218 may also be used to store information. For example, processor module 214, including memory 218, may be used as cache memory for the system, providing temporary storage for recent sensor data received from the sensors. In some exemplary implementations, the memory includes read-only memory (ROM), random-access memory (RAM), dynamic RAM, static RAM, non-static RAM, easily erasable programmable read-only memory (EEPROM), rewritable It may also include storage components such as possible ROM, flash memory, and the like.

A data storage device 220 may be coupled to the processor module 214 and may be configured to store various sensor information. In some exemplary implementations, data storage device 220 stores one or more days of continuous analyte sensor data. For example, the data storage may include 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, and/or 30 Days (or more) of continuous analyte sensor data may be stored. The stored sensor information may include one or more of the following. Timestamps, raw sensor data (one or more raw analyte concentration values), calibrated data, filtered data, transformed sensor data, and/or any other displayable sensor information , calibration information (eg, reference BG values and/or previous calibration information), sensor diagnostic information, etc.

User interface 222 may include various controls such as one or more buttons 224, a liquid crystal display (LCD) 226, a vibrator 228, an audio converter (eg, speaker) 230, a backlight (not shown), and/or the like. may include an interface. Components including user interface 222 may provide controls for interacting with a user (eg, host). One or more buttons 224 may be, for example, toggle, menu selection, option selection, state selection, yes/no responses to on-screen questions, "turn off" functionality (eg, for alarms), "acknowledged." It may enable functions (eg, for alarms), resets, and/or the like. LCD 226 may provide a user with visual data output, for example. Audio transducer 230 (eg, speaker) may provide audible signals in response to certain alarm triggers, such as current and/or predicted hyperglycemic and hypoglycemic conditions. In some example implementations, audible signals may be differentiated by timbre, volume, duty cycle, pattern, duration, and/or the like. In some exemplary implementations, the audible signal is generated by pressing one or more buttons 224 on sensor electronics 12 and/or by pressing a button or selection on a display device (e.g., key fob, cell phone, and/or or the like) to be muted (eg, perceived or turned off) by signaling the sensor electronics 12 .

Although audio and vibration alerts are described with respect to FIG. 2, other alert mechanisms can be used as well. For example, in some exemplary implementations, tactile alerts are provided that include poking mechanisms configured to "poke" or physically contact a patient in response to one or more alert conditions. be done.

Another battery 234 may be operatively connected to processor module 214 (and possibly other components of sensor electronics 12 ) to provide the necessary power for sensor electronics 12 . In some exemplary implementations, the battery is a lithium manganese dioxide battery, although any suitably sized and powered battery can be used (e.g., AAA, nickel cadmium, zinc carbon, alkaline, Lithium, Nickel Metal Hydride, Lithium Ion, Zinc Air, Zinc Mercury Oxide, Silver Zinc, or Sealed). In some example implementations, the battery is rechargeable. In some example implementations, multiple batteries may be used to power the system. In still other implementations, the receiver may be transcutaneously powered via, for example, inductive coupling.

Battery charger and/or regulator 236 may be configured to receive energy from an internal and/or external charger. In some exemplary implementations, the battery regulator (or balancer) 236 bleeds off excess charge current so that all cells or batteries within the sensor electronics 12 overcharge other cells or batteries. Regulates the recharging process by allowing the battery to fully charge without In some exemplary implementations, the battery(s) 234 are configured to be charged via an inductive and/or wireless charging pad, although any other charging and/or power mechanism may be used as well. can be used.

One or more communication ports 238, also referred to as external connector(s), may be provided to allow communication with other devices, e.g. 12 may be provided to enable communication with systems that are separate from or integral with. For example, communication ports include serial (e.g., Universal Serial Bus or "USB") communication ports that allow communication with another computer system (e.g., a PC, personal digital assistant or "PDA," server, etc.). may In some exemplary implementations, the sensor electronics 12 transmit historical data to a PC or other computing device (e.g., the analyte processors disclosed herein) for retrospective analysis by the patient and/or physician. ). As another example of data transmission, factory information may also be transmitted to the algorithm from sensors or from cloud data sources.

In some continuous analyte sensor systems, the on-skin portion of the sensor electronics may be simplified to minimize the complexity and/or size of the on-skin electronics, e.g. data, calibrated data, and/or filtered data only to a display device configured to perform the calibration and other algorithms required to display the sensor data. However, sensor electronics 12 (e.g., via processor module 214) may be implemented to execute predictive algorithms used to generate transformed sensor data and/or displayable sensor information. often, for example, assessing clinical acceptability of reference and/or sensor data, assessing calibration data for best calibration based on inclusion criteria, assessing quality of calibration, predicting analyte values. , comparing time-corresponding estimated analyte values, analyzing variations in estimated analyte values, assessing sensor and/or sensor data stability, detecting signal artifacts (noise), replacing signal artifacts, Determine rates and/or trends of sensor data, perform dynamic and intelligent analyte value estimation, perform diagnostics on sensors and/or sensor data, set modes of operation, evaluate data for the above and/or perform the like.

Although separate data storage and program memory are shown in FIG. 2, various configurations can be used as well. For example, one or more memories may be used to provide storage space to support data processing and storage requirements in sensor electronics 12 .

Calibration While some continuous glucose sensors rely on (and estimate the accuracy of) BG values and/or information obtained from the factory for calibration, the disclosed embodiments rely on real-time information (e.g., some In some implementations, only the sensor data itself) is used to determine aspects of calibration and to calibrate accordingly.

In some cases, analyte sensor calibration may use a priori calibration distribution information. For example, in some embodiments, a priori calibration distribution information or code is received as information from a previous calibration and/or sensor session (e.g., same sensor system installed) and stored in memory. , encoded at the factory (e.g., as part of factory default settings) on barcodes on packaging, transmitted from networks of cloud or remote servers, coded by caregivers or users, and from laboratory testing. Based on the result, it may be received from another sensor system or electronic device and/or the like may be performed.

As used herein, a priori information includes information obtained prior to a particular calibration. For example, from previous calibrations of a particular sensor session (e.g., feedback from previous calibration(s)), to information obtained prior to sensor insertion (e.g., factory information from in vitro testing or already embedded analysis data obtained from concentration sensors, e.g., sensors from the same manufacturing lot of sensors and/or sensors from one or more different lots), prior in vivo testing of similar sensors on the same host, and/or different Previous in vivo testing of similar sensors on the host. Calibration information includes information useful for calibrating a continuous glucose sensor, sensitivity (m), change in sensitivity (Δdm/dt) (which may also be referred to as drift in sensitivity), rate of change in sensitivity (ddm/ddt) , baseline/intercept (b), baseline change (Δdb/dt), baseline rate of change (ddb/ddt), baseline and/or sensitivity profile associated with the sensor (i.e., change over time), Linearity, response time, relationship between sensor characteristics (e.g., relationship between sensitivity and baseline), or US Patent Publication No. 2012/0265035-A1, which is incorporated herein by reference in its entirety. Specific stimulus signal outputs (e.g., outputs indicative of sensor impedance, capacitance, or other electrical or chemical properties) and sensor sensitivity or temperature (e.g., determined from previous in vivo and/or ex vivo studies), as described. ), sensor data obtained from an already implanted analyte concentration sensor, calibration code(s) associated with the sensor being calibrated, patient-specific relationship between sensor and sensitivity , baseline, drift, impedance, impedance/temperature relationship (e.g., determined from previous studies of the patient or other patients with characteristics in common with the patient), sensor implantation site (abdomen, arm, etc.), and/or or specific relationships (different sites may have different blood vessel densities), etc., but not limited to these. Distribution information includes ranges, distribution functions, distribution parameters (mean, standard deviation, skewness, etc.), general functions, statistical distributions, profiles, or the like representing multiple possible values for calibration information. When integrated, the a priori calibration distribution information is the range(s) or distribution(s) of values (e.g., , describing their associated probabilities, probability density functions, likelihoods, or frequencies of occurrence).

For example, in some embodiments, the a priori calibration distribution information is a probability distribution or sensitivity-related information for sensitivity (m) and baseline (b) or baseline-related information, e.g., based on sensor type. include. As noted above, prior distributions of sensitivity and/or baseline may be obtained from the factory (eg, from in vitro or in vivo testing of representative sensors) or from previous calibrations.

As previously mentioned, analyte sensors generally include electrodes for monitoring changes in current in either co-reactants or products to determine analyte concentration, eg, glucose concentration. In one example, sensor data includes digital data (eg, voltage or amperage) in "counts" converted by an A/D converter from an analog signal. Calibration is the process of determining the relationship between the measured sensor signal in counts and the analyte concentration in clinical units. For example, calibration allows a given sensor measurement in counts to be related to a measured analyte concentration value in milligrams per deciliter, for example. Referring to graph 10 of FIG. 3, this relationship is generally a linear relationship of the form y=mx+b, where "y" is the sensor signal in counts (y-axis 12) and "x" is the clinical value of the analyte concentration (axis 14) and "m" is the sensor sensitivity with units of [counts/(mg/dL)]. A line 16 is shown, the slope of which is called the sensor sensitivity. "b" (see line segment 15) is the baseline sensor signal, which can be considered or reduced to generally zero or near zero for advanced sensors. In any case, often the baseline can be estimated to be small in a predictable way, or can be compensated for. In some implementations, a constant background signal is found and such signal is modeled by y=m(x+c), where c is the glucose offset between the sensor site and blood glucose. .

Once line 16 is determined, the system can convert the measured counts (or amperages, eg, picoamperes, as described above) to clinical values of analyte concentration.

However, the values of m and b differ between sensors and require determination. Additionally, the slope value m is not necessarily constant. For example, and with reference to FIG. 4, the value m is _seen to change over time within a session from an initial sensitivity value _m0 to a final sensitivity value mF. The rate of change is considered greatest in the first few days of use and this rate of change is called _mR .

The slope is a function of in vivo time for many reasons. In particular, with respect to initial changes in calibration, such changes are often due to the sensor membrane "settling" into the in vivo environment and achieving equilibrium with that environment. Sensors are generally calibrated in vitro or on the bench, and efforts are made to make the in vitro environment as close as possible to the in vivo environment, but differences are still evident, and the in vivo environment itself varies between users. Change. Additionally, sensors may differ due to differences in sterilization or shelf life/storage conditions. Changes in calibration that occur later in the session are often due to changes in the tissue surrounding the sensor, eg, biofilm build-up on the sensor.

Whatever the cause, certain effects of variability have been measured and determined. For example, variability in final sensitivity _mF is known to be the largest contributor to overall sensor inaccuracy. Similarly, the variability and physiology of the initial sensitivity _m0 is known to be the largest contributor to day one sensor inaccuracy.

Because of the variability described above, the initial step of calibration is to determine and use seed values for one or more calibration parameters until additional data is obtained to adjust the seed values to more precise seed values. include.

Once calibration is achieved, the sensor and analyte concentration measurement system may be used to accurately determine the clinical value of analyte concentration in the user. The sensor and system can then provide other sensor behavioral distinctions, including determination of sensitivity error and drift, as described in more detail below.

The most common current method of calibration is through the use of an external blood glucose meter. Such a calibration method is commonly referred to as "fingerstick calibration" and is a familiar and common part of life for many diabetics. This technique has the advantage of not requiring significant factory information and furthermore offers a low risk of outliers. The disadvantage is that it requires significant user involvement, as well as knowledge of certain other factory calibration information required for proper calibration. Values from such meters can be used to calibrate indwelling analyte concentration meters, since measurements from such meters are trusted once the meters themselves have been calibrated. Although the sensor sensitivity changes over time as seen in FIG. 4, the changing sensitivity is not important if the user is willing to perform many external calibrations.

However, users are generally unwilling to perform many such calibrations and often do not, for example, for type II diabetics, pre-diabetics, or even non-diabetics. The additional precision provided by is not strictly required. For example, it may be sufficient for the user to know within which range they fall rather than the exact analyte concentration values. In another implementation, the data may be provided with an associated confidence interval to let the user know how much confidence to place in the displayed data.

Therefore, efforts have been made to reduce the number of calibrations. Nevertheless, many current CGM systems still require blood glucose levels to be used at least for initial calibration, which is often also required at the time of dosing. The present systems and methods according to present principles relate, in part, to methods of reducing or eliminating such required calibration.

One simple and convenient way to provide some level of calibration is through the use of calibration information on related sensors. Even if calibration information is approximate, such information may still be sufficient for use by certain patient groups. For example, and with reference to flow chart 18 of FIG. 5, if the factory calibration is known for one or more sensors in a manufacturing lot, this information is provided to other sensors in the manufacturing lot that have not yet been used on a patient. (step 20). This step is used to identify the manufacturing lot (and therefore details) of the sensor, as the code is often linked together during insertion into a patient as part of a CGM system. Therefore, it is often called providing a "code" to the transmitter. The transmitter may then identify the manufacturing lot from the code and apply the appropriate calibration parameters according to a lookup table or other technique. However, the code is not only a transmitter, but any device in which counts can be converted to clinical units, e.g., dedicated receivers, off-the-shelf devices that can be used to receive and display analyte concentrations, e.g., smart phones, It should be understood that a tablet computer or the like may also be provided. Additionally, the code may not be a code in the typical sense and may simply provide an arbitrary identifier to any device requiring it for calibration purposes.

Referring again to FIG. 5, a method for accomplishing the step of providing factory information about the sensor to a transmitter or other electronic device is described (step 20). Some of these methods constitute options for providing calibration information without using code (step 22). Other methods include using a step of user data entry to provide the code (step 28). In some cases, the code may be entered without user input (step 24). Still other methods of providing the code may be used (step 26), and such additional methods are also described below. Details of these methods are now described.

Referring to flow chart 30 of FIG. 6, an embodiment of providing factory calibration information in the absence of code (step 22) is described. A first method is to use a representative value, such as the mean or median value of manufacturing lots, or other measured value, such as a range (step 34). This means that even if the mean of a manufacturing lot is known, or even that of different manufacturing lots manufactured using the same technology, the sensor calibration parameters will be similar and therefore the It can be assumed that it can be used as part of the calibration. Alternatively, predictable relationships can also be used to interpolate sensor calibration parameters, eg, to bundle sensor lots.

As another example, impedance measurements may be used in determining calibration parameters (step 40).

Calibration may also be performed using information about previous calibrations (step 38). For example, if a user has just switched a sensor that has been calibrated and was measuring the user's glucose concentration at, say, 120 mg/dL, a proper measurement of the newly inserted sensor will indicate that the user's glucose is again at 120 mg/dL. It can be estimated that it should be such that In some cases, if predicted glucose values have been determined, the predicted values may be used for newly inserted sensors. Even if a short period of time has passed between the last reading of the old sensor and the new sensor reading, the recognition of the physiologically possible glucose change is critical to what the new reading could be and therefore the calibration of the new sensor. It imposes limits on what the parameters can be.

In yet another variation, various self-calibration algorithms may be used to self-calibrate the CGM system (step 36). In this sense, the CGM system may be referred to as becoming "self-aware." For example, the CGM system may use the previous session's steady state value or the previous session's slow moving average, e.g. May be seeded. The CGM system may also be seeded with A1C values, if available. Various inferences may also be made as appropriate. The seeded average value can be represented by a distribution, techniques for which are also described in more detail below.

FIG. 7 describes a system in which a code can be provided from the sensor to the transmitter without a step of user input (step 24). Again, although language is used here that refers to providing a code to a transmitter, the code may include a dedicated receiver, smart phone, tablet computer, follower device, or other computing environment for the transmitter and the signal. Note that it will be appreciated that various devices in communication may be similarly provided.

In the implementation of diagram 42 of FIG. 7, code or the like is provided to the transmitter, but without significant user involvement. For example, a degree of coding may be achieved by sending manufacturing lots of sensors to different markets (step 46). In one implementation, sensors with similar codes may be sent to different geographic locations (step 48). For example, a sensor sent to a particular geographic area may be from a similar or the same manufacturing lot, and when it is inserted and initially communicated with the network, calibration parameters known for that lot are , thus providing an instant degree of calibration based on geographic location. Geolocation information may be used to identify the location, which in turn may be used to identify or classify the sensor.

In the same way, sensors of similar lots may be grouped by product, so that different codes are then associated with different products (step 54). For example, a first product may have a first code associated with it, all of the sensors for that product may be manufactured in the same or similar manner, and the sensors associated with a particular product may resulting in little inter-manufacturing variability. In this case, once the product is identified, the associated sensor calibration parameters can be uniquely determined, at least as an average.

In another implementation, without particular consideration of geographic location or product, a particular group of sensors with a particular code may be shipped with a code specifically associated with the corresponding user's transmitter. (step 56). In this case, once the calibration parameters are known for one member of the sent group, and such parameters may be known long before the group is sent, the calibration parameters for the rest of the group are also known. is.

RFID technology may also be used to identify manufacturing lots of sensors (step 58). For example, a small RFID chip may be located at the base of the sensor and read by the transmitter when the sensor and transmitter are coupled (step 60). In another implementation, the RFID may be read by a receiver or smart phone or other device (step 62). Alternatively, the RFID device may be located on the applicator and the transmitter may read the identification information (and thus calibration information) again when the applicator is used to attach the sensor to the patient. .

In yet another implementation, Near Field Communication (NFC) may be used on the packaging or any other component of the system to communicate identification information (step 66).

Other types of communication schemes may be used to communicate information from the sensor to the transmitter. For example, a mechanical sensor on the transmitter may enable communication of the transmitter's coded information, e.g., bumps, vertical pins, mechanical system sensing directions, to a base or other mechanical element readable by the transmitter. (step 72). A magnetic sensor may be used for the same purpose (step 78), and in the same way an optical reader on the transmitter reads, for example, barcodes or QR codes as well as other identifying marks or colors. (step 74). A resistive sensor or other sensor that detects the state of the connection may also be used (step 76). For example, sensors of different codes may have corresponding different lengths. The sensor may have multiple contact pads on which the sensor is aligned. Different code sensors may have different resistances and the measurement may determine the code.

FIG. 8 shows a diagram 80 illustrating the method of code communication (step 28) using user data input. Perhaps most simply, the code may be provided to the user at the time of purchase, which is simply manually entered into the receiver, smart phone, or other device with a UI that allows data entry (step 84). . For example, a user may enter text, numbers, colors, and the like. The sensor may also be shipped with a card, e.g., a SIM card, which may be inserted into the receiver (step 90) to allow calibration information to be communicated, although the user may enter a manual code. No need to enter. The transmitter may be equipped with a switch system (step 88) and the user may adjust the position of the switch on the transmitter according to the received sensor instructions. For example, the transmitter switch may be a 4-position switch or a DIP switch, and with appropriate adjustment the user may provide the code associated with the sensor to the transmitter. The receiver or smart phone may also be able to scan labels associated with the sensor via an integrated camera or barcode reader, allowing information to be communicated that way. Scans may include bar labels, QR codes, and the like.

One variation is described below in FIGS. 9-11. This implementation uses human data from the field to improve or enable factory calibration. More specifically, factory calibration parameters (eg sensitivity and baseline over time) are often best identified using human data. Bench data correlates with human data, but the correlation is still not perfect and there are often offsets in the correlation. Having access to the human data generated by each lot of sensors produced during manufacturing is generally the best data set for generating factory calibration information. Since factory calibration parameters can vary from lot to lot, characterizing each lot can be advantageous when improvements are made.

Figures 9-11 illustrate how data collected on a person using a subset of many sensors is used to generate or adjust factory calibration numbers for the remainder of a lot. There are several arrangements for this method.

In position arrays, the calibrated sensors are sent to the market for use by patients. When such sensors are calibrated in a connected system, e.g., via blood glucose calibration techniques, or other calibration techniques, including calibration techniques that use only the CGM signal itself, the calibration information is available in the cloud or elsewhere. may be returned to the manufacturer through our Internet-based network. The information may be used to generate factory calibration settings for the rest of the manufacturing lot of sensors that were not sent to market, which may then be shipped.

In another implementation, there are initial factory calibration settings shipped with the product. Again, cloud or network information may be monitored and a determination made as to how closely the actual parameters match the initial factory calibration settings. Adjustments may then be made to the factory calibration settings of unshipped sensors based on this determined proximity. In this and previous implementations, sensor product releases may be staggered so that subsequent shipments improve accuracy. This implementation may even be performed even if all of the sensors are shipped, as adjustments can be performed over the network or over the cloud.

In more detail and with reference to FIG. 9, a factory 148 is shown with a manufacturing lot or batch 150 of sensors, which are generally made in the same (or very similar) manufacturing process. . A lot or batch 150 may be divided into a first portion 154 and a second portion 156 . A first portion 154 may temporarily remain at the factory 148 while a second portion 156 may be sent to a group of users 158 .

Referring now to Figure 10, the data from the second portion 156 informs the factory calibration of the first portion 154 and is processed by the factory 148 to convert it into a calibrated first portion 154'. may be used in If the first portion 154 has already been shipped, within the group of users 158, the first portion may be calibrated before or after insertion, indicated as first portion 154''. Calibration of the first part following shipment may be performed as described above by accessing a network or cloud resource for factory calibration information, especially if it has been updated with data from sensors in the field. .

FIG. 11 is a flowchart 160 illustrating the above method. First, a manufacturing lot or batch of sensors is manufactured at a factory under known and reproducible conditions (step 162). A lot or batch is divided into at least two portions (step 164). Although two parts are described here for convenience, a manufacturing lot may be split into any number of parts for staggered release.

In this example, the second portion is sent to the user (step 166). The second portion is then calibrated (step 168), which may be done in a known manner using, for example, a priori information, bench calibration values, user data, fingerstick calibration, etc. good. Calibration may also be performed using the techniques disclosed herein.

Calibration information from the second portion may then be sent to the factory (step 170). A calibration of the first portion of the sensor may then be generated or adjusted based on data from the second portion (step 172). That is, if a factory calibration has been generated for the first part, it may be adjusted as necessary. If a factory calibration has not been generated, the received data from the second portion may be used to inform the calibration of the first portion, such as the average of the determined sensitivities from the field. . The adjustment or generation may be done at the factory (step 174), or it may be done following shipment, before or after insertion into the patient (step 176), the transmitter, receiver, or other monitoring devices, such as smartphones, are in network communication with servers or other network resources operated by factory 148 .

Once the sensor is inserted into the user and the initial calibration is complete, any subsequent calibrations are referred to as "ongoing" or "ongoing" calibrations. Schematically, this is shown in diagram 102 of FIG. A user, patient, or host 112 has an indwelling sensor 114 connected to a transmitter 115 . Transmitters are often used multiple times for different sensors. In other cases, the transmitter may be disposable.

Transmitter 115 enables communication of signals measured by sensor 114 to a receiver or device such as dedicated device 104 or smart phone 108 . Receiver 104 is shown with display 106 and smart phone 108 is shown with display 110 . Display 106 or 110 may be used to show the user clinical value of analyte concentration, eg, glucose concentration. In doing so, they rely on the above-mentioned relationship that the measured current or count is related to the clinical value of the analyte concentration by a linear relationship with a slope representing the sensitivity.

Systems and methods according to the present principles describe the development or determination of this linear relationship based in large part or exclusively on the characteristics of the sensor signal itself, and in some implementations, as previous systems have relied on an external Data agnostic. In addition, such "self-aware" systems with "self" or "automatic" calibration not only allow subsequent analyte concentrations to be measured more accurately, but also retroactively correct the results of previous measurements. can also be used to In this way, when such a system is displayed on a display, such as display 106 or display 110, it more accurately conveys measured data. In other words, retrospective processing may be used to correct or modify previous calibrations and even update measured data therefrom. In this way, if the display shows historical data as well as current data, at least the historical data has been updated, i.e. its display has changed, and is known or more reliably known than the previous calibration. Reflect calibration parameters.

The method is illustrated by flow chart 116 of FIG. 13, where the first step is receiving or determining a seed value (step 118), which may be received or determined using, for example, the initial calibration procedure described above. . The seed value may then be used to determine calibration (step 120). For example, if the received calibration parameter is a particular value of slope or sensitivity m, it may be used to correlate measured counts to clinical values of analyte concentration, their measured It may be used to immediately begin notifying the user of analyte concentrations, eg, glucose measurements. That is, the analyte may be measured using the sensor (step 122) and the measurement may be displayed to the user based at least in part on the seed value received in step 118 (step 124). .

In some cases, a change in calibration occurs (step 126), which may be detected in a variety of ways, including those described below. Calibration, specifically calibration parameters including sensitivity and baseline, may then be adjusted (step 128). After updating the calibration parameters, the display may be updated (step 130).

As described above, updating the display may refer not only to adjusting the currently measured value of the analyte concentration, but also to recalculating the historical value and changing the display based on the adjusted calibration. . For example, the sensitivity may have been "seeded" by an initial value, but following receipt of the data it may be determined that it is actually 10% lower than the initial seed value. In this case, it is not only the ongoing displayed analyte value that is adjusted, but in one implementation the historical value may also be adjusted to reflect the updated sensitivity. This example shows a situation where the seed value is updated with the measured value. In some cases, already determined values (determined by seeds or measurements) may be updated with later determined values. This situation can arise, for example, when the sensor calibration parameters "drift". For example, if it is determined that the calibration of the sensor has drifted, changes may be made to the calibration parameters so that the receiver, smartphone, or other monitoring device continues to display accurate values of analyte concentration. In one implementation, certain historical values, i.e. values measured following drift, may be updated in the display if it can be determined when drift occurred, while other historical values, e.g. , the values measured before the drift need not be updated.

In one implementation, if the determined seed value and the initial seed value are close, eg, within 10%, the initial seed value (or other calibration parameter) may simply be adjusted accordingly. However, if the values are further apart, the user may be prompted for optional fingerstick intervention, for example.

Figures 14 and 15 are graphs showing analyte concentration over time before (14) and after (15) the change in sensitivity. Specifically, FIG. 14 shows a graph 132 in which a plot 138 of analyte concentration over time is shown. Axis 134 represents analyte concentration values and axis 136 represents time. Following the change in sensitivity, the graph becomes graph 140 with transformed historical analyte values 146 . A change in sensitivity may be considered an updated sensitivity or an updated seed value, depending on the implementation. Calibration using sensor signal characteristics including, but not limited to, the mean sensor signal, standard deviation of the sensor signal, or CV (coefficient of variation), or interquartile range, or other higher-order or rank statistics. Several other methods of adjustment can be used.

More specifically, and in contrast to previous efforts, preferred embodiments process substantially real-time graphical representations of glucose data (e.g., trend graphs representing glucose concentrations over the previous number of minutes or hours). Systems and methods are described for periodically or substantially continuously post-processing (e.g., updating) with the obtained data, wherein the data is responsive to calibration updates, e.g., sensor drift, It is being processed as a result of a system error, etc.

Referring specifically to block 137, the analyte concentration measurement system 135 shown in FIG. 16, a sensor data receiving module, also referred to as a sensor data module, or processor module, receives one or more time-spaced sensor data points. Receive sensor data (eg, a data stream), including: In some embodiments, the data stream is stored in the sensor for further processing, and in some alternative embodiments, the sensor may be in wired or wireless communication with the sensor, such as a receiver such as a smart phone. Or periodically send data streams to other monitoring devices. In some embodiments, raw and/or filtered data is stored in the sensor and/or transmitted and stored in the receiver.

At block 139, the processor module is configured to process sensor data in various ways. The processor module may also be used in combination with the calibration module 143 to determine if a calibration change has occurred, as described above and in more detail below. More specifically, at block 143, the calibration module uses the data in the data stream to detect changes in calibration, and more specifically in sensitivity.

At block 141, the output module provides output to the user via a user interface (not shown). The output represents an estimated glucose value, which is determined by converting sensor data into meaningful clinical glucose values. The user output may be, for example, in the form of numerical estimated glucose values, directional trends in glucose concentrations, and/or graphical representations of estimated glucose data over time. Other displays of estimated glucose values are also conceivable, such as audio and tactile. In some embodiments, the output module provides both "real-time" glucose values (e.g., numerical values representing the most recent measured glucose values), as well as graphical representations of processed and/or post-processed sensor data. indicate.

In one embodiment, the estimated glucose value is represented numerically. In another exemplary embodiment, the user interface graphs estimated glucose data trends over a predetermined time period (eg, 1, 3, and 9 hours, respectively). In alternate embodiments, other time periods may be represented. In alternate embodiments, photographs, animations, charts, graphs, ranges of values, and numerical data may be selectively displayed.

The processor module may be further configured to perform post-processing steps, such as cycling the displayed graphical representation of the data corresponding to the time period according to the received data, e.g., more recently received data. It may be configured to post-process (eg, update) continuously or substantially continuously. For example, glucose trend information (eg, for trend graphs for the previous 1, 3, or 9 hours) may be updated to better represent actual glucose values considering newly determined calibration values. In some embodiments, the post-processing module performs the post-processing module every few seconds, minutes, hours, days, or anywhere in between and/or when requested by the user (e.g., displaying a 3-hour trend graph). In response to a button activation such as a request), post-process segments of data (eg, 1, 3, or 9 hour trend graph data).

Generally, post-processing is performed by a processor module (e.g., in a handheld receiver unit) on "recent" sensor data (e.g., data containing time points within the last few minutes or hours). After its initial display, and for what is commonly referred to in the art as "retrospective analysis" (e.g., display of sensor data for physician analysis, etc.), intermittent, periodic analysis performed retrospectively as opposed to retroactively or continuously)). Post-processing includes recalibrating sensor data (e.g., to better match reference values), filling data gaps (e.g., data excluded due to noise or other problems), and smoothing sensor data. (filtering), compensating for time lags in sensor data, etc. can be included. Preferably, post-processed data is delivered in "real-time" (e.g., including recent data within the last few minutes or hours) and on a personal handheld unit (e.g., receiver or smart phone 1, 3 , and on a 9-hour trend graph) and automatically (e.g., periodically, intermittently) when new or additional information is available (e.g., new reference data, new sensor data, etc.) or continuously), or selectively (eg, in response to a request) (post-processed). In some alternative embodiments, post-processing may be triggered according to the duration of the calibration episode change, e.g., data associated with a calibration event change that extends beyond about 30 minutes is may be processed and/or displayed separately from the first 30 minutes of data.

In one exemplary embodiment, the processor module filters the data stream to recalculate data over a previous time period and periodically or substantially displays a graphical representation of the recalculated data over that time period. Continuous display (eg trend graph). In another exemplary embodiment, the processor module adjusts the data for time lag (e.g., removes time lag induced by real-time filtering) from data over a previous time period, and adjusts the time lag adjusted data over that time period. display a graphical representation of the (e.g., trend graph). In another exemplary embodiment, the processor module scans one or more sensors over a moving window (e.g., including time points before and after one or more sensor data points) for data over a previous time period. Algorithmically smooth the data points and display updated, averaged, or graphical representations of the averaged data over time (eg, trend graphs).

In some embodiments, the processor module is configured to filter sensor data and display a graphical representation of the filtered sensor data in response to determining the initiation of a change in calibration event. In some embodiments, the processor module is configured to display a graphical representation of unfiltered sensor data (eg, raw data) in response to determining the end of the change in calibration event. In some embodiments, the processor module is configured to display an unfiltered graphical representation of the sensor data except when a change in calibration event is determined. The adaptive filtering described herein, including selective filtering during calibration event changes, increases the accuracy of displayed data, reduces the display of noisy data, and/or reduces data gaps, and /or has been found to shut off earlier than conventional sensors.

Calibration Routines As noted above, it is desirable to provide more convenient calibration routines for users, particularly Type II users, or those who use the system for weight loss optimization and/or sports and fitness optimization.

One way to reduce the need for user-based calibration is to use a more robust factory calibration, and certain details regarding the methods associated with factory calibration have been published as US 2014-0278189-A1, See USSN 13/827,119 filed March 14, 2013 and USSN 62/053,733 filed September 22, 2014, both of which are owned by the applicant of the present application and incorporated by reference therein. is incorporated herein in its entirety.

Other techniques may also be used to facilitate calibration requirements. For example, referring to flowchart 145 of FIG. 17, if the previous sensor session showed generally reliable results (step 147), the same calibration parameters from the previous session may simply be used for the new sensor session (step 147). step 149). Specifically, calibration parameters from old sensor sessions can be retrieved in many ways, e.g., if the calibration parameters are stored on the sensor electronics, by using glucose signal transmitter technology, or if the calibration parameters are stored on a monitoring device. For example, if written to a smart phone, it may be sent to the new sensor session by passing the calibration state variables to the new session. This technique can be particularly useful if the sensors are more or less related, eg, from the same batch of the same type, the same package, and the like.

This technique is not necessarily limited to using only a single previous session, eg, returning to only one session. For example, recurring patterns, eg historical patterns, can be learned by analysis of several previous sensor sessions. For example, the system may learn that the user is in the habit of eating pizza on Fridays and has been doing so for the last 7 sessions, so the algorithm will not treat such events as outliers. can learn to The patient's habits, eg, the patient's preference for eating many small meals versus only a few large meals, can also be learned. Failure modes can also be learned, for example, if a patient is predisposed to a particular failure mode due to a particular way of mounting and/or using their device. As discussed below in connection with FIG. 18, certain glucose trace characteristics that make up a repetitive event may advantageously be learned from previous sensor sessions and are often common events. To distinguish from outliers, multiple previous sensor sessions are required. Additionally, if other event data is available, such as diet or exercise data, correlations between such repetitive events and input diet or exercise data can be learned.

Referring now to Figure 18, a flowchart 178 for another method of calibration is shown. Specifically, certain glucose trace profiles are known to exhibit recurring events at known and recurring glucose values when they recur. By way of example, steady state values, certain trend values with certain slopes, etc. tend to be reproducible for a given patient. Such repetitive events tend to give rise to characteristic glucose trace features and/or patterns that are repetitive and detectable. For example, it is characteristic of many biological systems that analyte values are highly reproducible when they are not changing rapidly, ie, at steady state. In other words, if an analyte value is steady state for a first time period and then steady state for a second time period, the analyte value, e.g., concentration, is generally , is or is close to the same value. This concept can be used to calibrate the analyte sensor.

For example, if the user is steady-state with respect to the analyte, the value of the analyte can be measured and stored. When the users are again at steady state, their analyte values are likely to be the same as the previously measured values, so the sensors reading the analyte concentration values can be calibrated.

Different analytes can achieve different steady states in different ways. For example, uric acid concentrations change very little throughout a typical day. If the user has not exercised and eaten for several hours, their glucose levels may be at steady state. If the user has not exercised for several hours, their lactate value may be at steady state. Generally, many analyte values, including glucose, achieve a steady state if the biological system has not changed significantly over time. As noted, the steady state values are reproducible, especially for pre-diabetics or non-diabetics, and for humans using the system primarily for weight loss optimization or sports optimization. Therefore, a sensor that measures an analyte can be calibrated whenever a steady state is detected in the analyte value.

In some cases, the system may prompt the user to fast or exercise to achieve steady state, which can then be measured and subsequently used for such calibration. Additionally, the system detects a steady state but may prompt the user for verification, for example, by asking the user "Are you fasting?"

In some cases, the steady state value may be determined based on the user's demographics and thus does not require any measurements. For example, for non-diabetics, typical glucose levels can be about 80-100 mg/dL. In many cases, their value may be around 80 if the person is very non-diabetic, while their value may approach 100 if they are progressing to pre-diabetes. Thus, by providing the system with only certain information about the user, in particular the user does not have to determine the exact value, rather accuracy only to a certain range is sufficient, e.g. For certain applications, including when , hypoglycemia, or euglycemia, some degree of calibration may be allowed to be performed.

Besides steady state, other repeatable events that may be used include slopes, typical or similar post-meal responses, eg, post-breakfast responses if the user eats the same type of breakfast every day. Other repetitive events are, for example, daily low to high reading ranges, i.e., daily high and low ranges, certain types of excursions, certain types of temporal patterns, decay rates, and slopes. , rate of change, etc.

As a particular example, if a user has a characteristic breakfast, and the characteristic breakfast results in a characteristic postprandial glucose trend, if a change in trend occurs, then at least to some extent, a change in sensitivity of the sensor has occurred. , it can be estimated that the sensor needs to be recalibrated.

Additional data may also be used to aid calibration. For example, if the user is diabetic and measures their blood sugar in some way several times a day, such a value can be used as the calibration value. This is especially true if their blood sugar levels change significantly. In this case, if the system detects a local steady state, the system may prompt the user to measure and enter a blood glucose level in order to correlate the value with steady state.

Historical data, in some cases, to determine steady-state calibration values, e.g., previous data from blood records, data from previous sessions, extrapolations from measurements such as A1C that track long-term glucose averages, etc. can be used for

In some cases it is not necessary to focus on a particular value. A determination that the user is within a range of values may be sufficient for Type II users. Ranges may be determined and used to provide users with information as to whether goals are being met, or information about programs they are receiving.

The use of steady-state estimates can be used not only for initial calibration, but also for update calibration. That is, whenever the system is assumed to be in a subsequent steady state, the values measured during the subsequent steady state can be assumed to be the values originally determined for the user. Other calculations may also be used, including using weighted averages, slow moving averages (see below), and the like.

Referring to FIG. 18, the method of flowchart 178 allows calibration of glucose concentration or other analyte values using information generated directly from the device, ie, the analyte concentration signal itself.

Thus, referring to flow chart 178, the first step in a calibration routine according to these principles is to detect that the system is at steady state (step 180), e.g. .

Generally, a steady state can be detected without any action by the user (step 182). However, in some cases, steady state is achieved by waiting a predetermined period of time after an event, for example by waiting a predetermined period of time following a diet or exercise routine (step 186). You may be prompted to enable (step 184). The system or routine may, for example, request or prompt the user for information regarding fasting, for example prompting the user to enter the duration since the user's last meal or related parameters. For glucose, the routine may be configured to seek a low rate of change, or a rate of change less than a predetermined value, eg, less than 0.25 mg/dL per minute. In some cases, if calibration has not yet occurred, the rate of change may be based on counts, or microamps, or other "raw" signal values. The rate of change may be determined by derivative calculations.

Once steady state is detected, calibration parameters such as sensitivity may be determined (step 188) based on the measured counts (or current) at steady state and the known steady state glucose value, which is the Or it may be an estimate. Another steady state can then be detected (step 192), and if drift is occurring, the second steady state is associated with a different count number. A different number of counts may be used to determine the degree of drift, with the system adding new (second) counts measured at a new (second) steady state along with the already known steady state glucose value. It may be recalibrated using numbers (step 194).

In some cases, if the change is substantial, e.g., exceeds a predetermined threshold, the user may be prompted for additional data, e.g., fingerstick, exercise or diet related data (step 196). . Differences may also be used to determine the cause of drift (step 198).

Specific examples are described here. A continuous blood glucose monitor may be used for patients who do not have diabetes. In non-diabetic patients, their glucose levels are typically in the range of 70-90 mg/dL. Their glucose levels fall outside of this range only after meals or during strenuous exercise, and in some cases such deviations can be detected by changes in sensor signals or through supplemental measurements using heart rate monitors or accelerometers. may be detected. During these glucose excursions, the current measured by the sensor changes rapidly, which can be easily detected by monitoring devices. The rate of change can be calculated, for example, using an FIR filter over the last 20 minutes of glucose values, but as the difference between the two glucose values at the beginning and end of the period divided by the duration of the period. It can range down to a simple rate of change that is defined. During fast rates of change, systems and methods according to the present principles can avoid such rapidly changing data for calibration when using steady state occurrences as repeatable events for calibration. , rather wait until the values have stabilized before performing a calibration event (as noted, certain calibration methods may utilize such; e.g., as noted above, in some cases Repeatable events that can be used for calibration may include temporal noise events or patterns, i.e. certain aspects of high change regions or peaks may be useful for non-steady-state repeatable event calibration, and many , transient events involving rapid or slow rates of change in analyte concentration values can form signal characteristics whose patterns are estimated and used as repetitive events). In one example, the absolute rate of change threshold may be set at 0.25 or 0.5 mg/dL/min over the last 25 minutes of glucose data. As noted above, uncalibrated units may also be used. Therefore, calibration may be prohibited if the absolute rate of change threshold is exceeded. In other implementations, different calibration values may be used, which may also be configured to depend on the direction of the rate of change, or the calibration value used may be a function of the rate of change itself. .

Referring to FIG. 19 and as described with respect to step 194 of FIG. 18, steady state may be used to update calibration values and to determine initial values. Moreover, the system can be used even when the calibration parameters change, for example when the sensor enzyme layer changes over time with use or when other drifts occur. For example, referring to graph 202 of FIG. 19, analyte values may be associated with calibrated or uncalibrated values at time t ₀ (axis 206), this steady state value being the Whenever there is steady state at that analyte value, it can be assumed to be reproducible. This knowledge can yield an initial calibration line 208, the slope of which is the sensor sensitivity when the axis 204 has units of counts. At subsequent times t1, which are also assumed to be steady _- state, knowledge of the same steady-state value allows subsequent calibration lines 210 to be drawn, thus even though the number of counts measured is different, thus , the system can be recalibrated. The degree of recalibration required can be very useful in determining the cause of drift and treatment.

Systems and methods according to the present principles can be most effective when the baseline or background signal is sufficiently stable and predictable (or eliminated through advanced membrane or sensor technology). A current is generated when the sensor is activated. If the baseline is small enough or estimated with sufficient accuracy, the remaining current may come from the analyte of interest. The algorithm may measure the current over a set period of time, and if the current is stable, e.g., within predetermined limits, the algorithm determines that glucose (or other analyte) has not changed and It can be estimated to be within a narrow range. The algorithm then uses, for example, a glucose value of about 80 mg/dL (or whatever is determined to be a typical value for the user) and changes it to The device may be automatically calibrated by correlating the current generated in . In one implementation, the glucose value was set at 100 mg/dL (it was repeated that it could vary depending on factors such as rate of change, duration of wear, time of day, user characteristics, e.g., state of progression to diabetes). ). The system and method may use a regression model to calculate the slope and baseline at two points. The first point is the current produced during stable glucose (and the approximated glucose level of a non-diabetic user) and the second point is zero glucose (estimate for background signal using). The slope of the line may be determined, for example, using the regression slope (estimated for counts/BG) and a weighted average of previous slope estimates. Following calibration, glucose data may be presented to the user. The baseline for this implementation was assumed to be zero, but different non-zero baseline values can be used. The calibration may also be updated periodically, eg, every few minutes or hours depending on the implementation.

The above technology is owned by the assignee of the present application and is incorporated herein by reference in its entirety, in US patent application number It may be used in conjunction with factory calibration information generated during manufacture of the device, as described in more detail in USSN 13/446,848, or it may also be used with externally generated glucose information. The time-varying sensitivity may be further accounted for by incorporating predetermined drift curves or other drift compensation techniques.

Systems and methods according to the present principles may further be used to use calibration information for one sensor to calibrate another, e.g., adjacent sensor, e.g., sensor under the same membrane. Such calibration may be performed when the drift parameter, if caused by the membrane, can be assumed to be the same for both sensors. For example, if both sensors, e.g., a glucose sensor and a lactate sensor, are under the same membrane layer, and one or more calibration parameters were determined ex vivo, the calibration parameters are said to have a similar relationship in vivo. can be estimated, so measurements of one can be used to determine the other. For example, if a lactate sensor has a known calibration offset (or other relationship, or scaling, or correlation coefficient) from a glucose sensor, as measured ex vivo and then in vivo, then the determination of the calibration for that glucose sensor. may be used to calibrate the lactate sensor. For example, if the calibration of the glucose sensor is considered to have drifted by 50%, the calibration of the lactate sensor can be estimated to have drifted by 50%. As a result, updating one or more calibration parameters of one sensor may result in updating one or more calibration parameters of the other sensor.

Further details of such aspects can be found in US patent application Ser. See published USSN 12/829,264, filed July 1, 2010, and [545PR], all of which are owned by the applicant of the present application and incorporated herein by reference in their entirety. be

Additionally, systems and methods according to the present principles may start using factory calibration information and then incorporate auto-calibration techniques over time to obtain more accurate glucose information. If the signal did not comply with the predetermined parameters or was outside the predetermined parameters, the system may request calibration values using known techniques, eg, SMBG or fingerstick calibration. The system and method may then incorporate this glucose information into the original parameters to adjust the setpoint, eg, from 100 mg/dL to a more appropriate and accurate value. That is, while the above techniques intended for use in certain applications can generally be configured to avoid the need for fingerstick calibration, if they are available, systems in accordance with the present principles and the method may be advantageously applied for calibration purposes or otherwise.

Systems and methods consistent with the present principles may be configured to determine a confidence level or range, which may change as the resolution or accuracy of the data changes. More specifically, the display may generate value and trend graphs, or it may show ranges or other UI elements. The range changes over time and may shrink or expand as confidence in accuracy changes. For example, during initial warm-up, factory calibration values may be utilized. However, the accuracy may not be as precise as having the additional information. During this time the display may show ranges rather than values.

A further feature of systems and methods according to the present principles is that they may request information when the user is setting up within the system and adjust which techniques to use according to that information. is. For example, the system may prompt the user to input whether the user has type I diabetes, type II diabetes, or is non-diabetic, and may select different techniques depending on the answer. good. Systems and methods according to the present principles may also ask if the user is interested in weight loss optimization, sports and fitness optimization, or other similar optimization routines, and adjust the algorithm accordingly. may The device may also be used in a "blind" mode for an extended period of time, eg, 14 days, and only receive blood glucose levels. These blood glucose values can be used to learn what the patient's resting blood glucose is, which can better guide the estimation of auto-calibrated steady-state blood glucose values. After an extended period of time, the user can then use the device in autocalibration mode.

Besides the use of properties of the steady-state value of the analyte to gather additional information, a "slow moving average", i.e., an average value measured over, for example, 1-3 days, may also be used, That is because such slow moving averages are also generally constant, especially over the use of a sensor session. Therefore, the variation can be used both qualitatively and quantitatively to detect and quantify drift. For example, the slow moving average values of glucose concentration are shown by graph 220 in FIG. 20 and the more schematic graph 212 in FIG. 21, with sensor counts shown on axis 214 against time on axis 216 . As can be seen, the slow moving _average G1 measured at time t1 can decrease to the slow moving _average G2 at time t2. The slow moving average can be used to quantify drift, as advanced sensors are highly selective to glucose and therefore sensitivity is the only contribution to changes in the slow moving average.

Details of the use of the slow moving average are detailed below, but it should be noted here that it need not have continuous periods. For example, a slow moving average may be measured by sampling common time periods over many days. For example, a slow moving average of the nighttime period may be measured, which may then consider only the period, eg, from 11pm to 7am. A slow moving average then constitutes an average of the data measured over several days, but only during this period. Other exemplary time periods during which such discrete or discontinuous slow moving averages may be measured may include, for example, postprandial periods. Such events can be time-based, where the user has very consistent timing of such events, or can be event-based, where events are recorded or tracked by one or more sensors, for example. obtain. For example, exercise events may be recorded by accelerometers and meal events may be recorded by detecting glucose spikes.

More specifically, instead of using daily averages, averages are measured that are specific to a qualitative or quantitative kind of period that is specific and important to the user, e.g. A time period may be used to make the frame unique. For example, the time frame may be considered "4 hours after meals". Period data can be used as measured over a number of days, but considers only that particular period, and thus only measures the variation from the average of the glucose response during that particular set of periods. In other words, the average value may be determined by summarizing and averaging all of the glucose values from individual time periods over several days. In this way, if drift is detected, it is defined with respect to the average value obtained by measuring the average over such similar time periods. For example, a patient's daily average may be 100, but their nighttime average may be 85, and their "regular" walking daytime average may be 120. Drift measurements may be made with respect to this defined "local" average. Exemplary time periods may include, for example, after dinner, from 9:00 am to noon, at sleep, and the like.

A slow moving average of calibrated or even uncalibrated values may be measured, but the slow moving average itself may require additional data to perform the initial calibration. For this reason, implementations may involve using other methods to determine the initial glucose value and, for example, obtaining a day's worth of data, from which an initial slow moving average is determined. may then be compared to subsequently measured slow moving averages to determine, for example, sensitivity drift corrections. Such systems and methods are significantly less expensive compared to previous systems, in contrast to SMBG calibration where the slow moving average can only be done as often as the user is willing to take readings. It can be particularly useful as it can be checked frequently, eg every 5 minutes.

Flowchart 222 of FIG. 22 illustrates one implementation of using a slow moving average. In a first step, an analyte value is measured using the sensor (step 224). A first slow moving average may then be determined (step 226). The analyte value may then continue to be measured (step 228), which may form the basis for the indicated value of analyte concentration (step 232), the indicated value being the initial sensitivity. Based on (or subsequent) values. A second and subsequent slow moving average may be determined (step 230), the duration of the slow moving average generally exceeding half a day, eg, 1-3 days. Slow moving averages may often be measured as often as requests, eg, every 5 minutes, every hour, and so on. In some implementations, the filter's time constant can be changed, for example, when the user has an actual maximum, so that the effect (user's high analyte concentration) is greater than the actual sensitivity of the filter. Does not represent drift or change. The use of slow moving averages may also be replaced by other forms of filtering such as ordinal statistical filtering or time domain filtering.

If the first and second values (or second or subsequent values, or indeed any set of values) of the slow moving average change, in one implementation drift may be presumed to have occurred, The calibration may be adjusted based on drift, eg, the difference between slow moving averages (step 234). The implementation described below considers other potential sources of slow moving average fluctuations. The display may be updated based on the recalibrated sensor (recalibrated based at least in part on the measured drift) (step 236). In some cases, the seed value (or other value used as a basis for display) may also be modified to reflect (and compensate for) drift (step 238).

Referring to flowchart 240 of FIG. 23A, recalibration (or other recalibration) based on data collected from changes in the slow moving average is also used in post-processing, where "already The display of historical values, which are defined as "displayed" values, may be updated. In a first step, an analyte concentration value may be measured by the sensor (step 242). The measured value may be displayed as a clinical value based on calibration, eg, based on previously determined sensitivity values (step 244). Sensitivity may vary based on (including based entirely on) sensor signal information (step 246), for example, based on slow moving averages or changes in steady state values. The display may then be updated based on the changes (step 248), specifically previously displayed values of the history are redisplayed based on the recalibration so that the historical values are more accurately represented. may Other forms of detecting sensitivity change (or drift) using signal characteristics (or features) are the CV (coefficient of variation), standard deviation, or interquartile range of the signal and the parameters of glucose to which the relationship corresponds. including.

Once a change is detected, it may be analyzed or "differentiated" to determine the cause and/or magnitude of the change. It is common to find changes in sensitivity due to drift, but it can also have other causes, including pump problems, such as blockage of the pump, or other problems, such as broken membranes. It is further desirable to distinguish these changes from actual changes in glucose levels. As a first step in at least this latter determination, the measured change in glucose level may be compared to a threshold for such change that is physiologically feasible. If the change is not physiologically realizable, the change can be attributed, at least in part, to drift or system anomalies.

Another way to distinguish signal drift behavior is by comparing a signal drift curve to known signal drift curves, specifically multiple envelopes of such curves. FIG. 4 shows one such curve, but for a given type of sensor there is an envelope of such a curve, ie a pattern of how the sensitivity varies. If the method the sensitivity is changing follows one of these curves, it can be inferred that the change is due to sensitivity drift rather than actual glucose values or system anomalies. Further details regarding sensitivity profile curves are found, for example, in "Systems And Methods For Processing Analyte Sensor Data," filed September 19, 2013, owned by the assignee of the present application, and incorporated herein by reference in its entirety. US Patent Application Serial No. 13/796,185.

If the sensitivity changes by moving to another sensitivity on a known sensitivity curve, known calibration parameters for that curve may be used in subsequent data analysis and display. If the sensitivity varies outside the limits of the known sensitivity curve, it can be inferred that the variation is due to system problems or anomalies, eg, errors or artifacts, as described above. However, in certain implementations, certain sensors may have sensitivity curves that are known to be in-band. A known failure mode may shift the sensitivity to another curve within the band or to another band, ie the sensitivity may shift to another known band of the curve with a known failure mode.

In this regard, it should be noted that in general a given type of sensor is functional to achieve the required goals up to a certain tolerance. For example, 80% of a lot's sensors may function as desired (see Figure 23B). The remaining 20%, on the other hand, may not follow the predicted sensitivity curve (see Figure 23C). A majority of these remaining sensors, eg, 75%, may follow known failure modes, which causes them to follow known alternative sensitivity curves, families of curves, or bands of curves. By identifying which of these sensors follow alternate sensitivity curves and adjusting the calibration of the sensors accordingly, sensor "failure" can be ameliorated in large part. For example, if the failure mode is such that 75% of all have signal values that tend to decrease in the same way, after determining the failure mode, "failure" adjusts the sensor reading "up". can be improved by Such aspects may be particularly important as sensor sessions become increasingly longer, eg, from 7-day sessions to 14-day sessions. In the failure mode shown in FIG. 23C, the decrease in sensitivity begins around day 8. The ability to detect and ameliorate such failure modes is quantifiable, especially as session sensor failure mode exits with longer sessions become increasingly well-characterized, even as session sensor failure mode initiation becomes increasingly well-characterized. This is particularly important because it remains difficult to

In some implementations, consideration of glucose signal variability combined with a slow moving average may be used to distinguish glucose signal fluctuations from fluctuations in sensor sensitivity. For example, if the slow moving average decreases but the variability remains the same or stays within a specified range or band, the slow moving average decrease is likely due to sensitivity changes. Alternatively, if the slow moving average decreases, but the variability also decreases, the cause of the decrease is likely real and actual change in glucose concentration.

Changes, variations, errors, artifacts, and other signal behavior that are not physiologically realizable can be the cause of various remedial actions by the system, some of which are mentioned above. For example, recalibration may be performed and the results propagated back toward historical data. The user may be prompted to provide fingerstick calibration. Part of the corrective action algorithm may be deciding whether to correct through recalibration or prompt for fingersticks or other calibration points. For example, if a signal is received that falls outside the limits of physiological plausibility, the user may be prompted for additional calibration points, eg, fingersticks. Alternatively, the user may be prompted to provide other types of additional information, such as entering data corresponding to recent exercise or diet or other recent changes in user behavior. As a specific example, if a slow moving average of a user's glucose concentration was 100 mg/dL for the first three days of the session and then jumped to 200 on the fourth day, systems and methods according to the present principles may prompt for fingerstick calibration. If the slow moving average on day 4 was 105 mg/dL, the system may increase or decrease or adjust the sensitivity accordingly, for example to bring the value down again to 100 mg/dL.

If fingerstick calibration is performed, the use of fingerstick calibration data may vary depending on device usage. For example, if the device is being used adjunctively, i.e., non-therapeutically (which is the case for many Type II users), if the fingerstick shows drift, use the sensor if the drift is not substantial. It may still be possible to do In many cases, the techniques described herein can be used to ameliorate the effects of drift and still allow display of accurate readings. If the device is used non-adjunctively, e.g., for type I patients using insulin, if the fingerstick exhibits drift, improvements or adjustments, e.g., recalibration, may be made more aggressively. Often, if it is not possible to do so, or if accurate sensor readings cannot be returned after recalibration, the user may be prompted to discontinue use of the sensor.

Pattern analysis may be performed to determine if the changes or variations are of a known type, eg, characteristic of known sensitivity changes. Pattern analysis may determine whether changes or variations meet criteria for certain behaviors, eg, exceed certain known thresholds. As mentioned above, daily behavior may be used in determining the slow moving average. Another day's data may be used if, after analysis, the system determines that the data for that day is unreliable. Data may be displayed in ranges or bands, or with confidence intervals or other indicators, rather than as high-precision numbers. If a slow moving average is used, the time constant of the slow moving filter may be adjusted to include or exclude shorter term fluctuations.

These aspects are summarized in flowchart 250 of FIG.

Referring first to FIG. 24, an analyte concentration value may be measured by the sensor (step 252). Such values are generally measured as current, eg, as amperes (picoamperes) and equivalently as counts. A slow moving average may be defined by measuring counts over an extended period of time, such as several hours, half a day, 24 hours, or 2-3 days. Because the sensor varies from unit to unit, the slow moving average generally becomes meaningful only once an initial period of time, eg, 24 hours, has elapsed. Therefore, the next step is to determine a first slow moving average over a first time period (step 254).

In one implementation, a value for initial use of the first apparent sensitivity can be assumed via a seed value as in the method above. A first apparent sensitivity may then be determined based on the first slow moving average (step 256). Specifically, if a first slow moving average is assumed, the first apparent sensitivity is based on the relationship between the assumed first slow moving average and the measured first slow moving average. may

A second slow moving average may then be determined over a second time period (step 258). Again, optionally, a second apparent sensitivity may be based thereon (step 260).

If the slow moving average or apparent sensitivity is deemed to change between the first period and the second period, remedial action may be required. Therefore, the next step is to determine if the change meets predetermined criteria (step 262). The predetermined criteria may include many factors, such as known sensitivity changes over time, the envelope of the sensitivity profile curve, physiologically likely changes, changes associated with errors such as pump malfunctions, etc. (step 270). For example, such a step may call for a determination as to whether the change is consistent with criteria associated with sensitivity drift, anomalies, or actual changes in mean glucose concentration values (step 264). If consistent with drift, eg, if it is determined that the sensitivity has drifted by comparison to a known sensitivity profile curve, correction may be made automatically. In either case, the sensitivity may be adjusted (step 268) based at least in part on the difference between the two slow moving averages, which is a quantitative measure of the degree of change or drift experienced by the sensor. This is to provide an index.

If the change is not consistent with drift, it may be determined if the change is consistent with other causes for which predetermined criteria exist. If the change does not match known behavior, e.g., does not match any predetermined criteria, the user may provide information to describe the change, e.g., dietary or exercise information, fingerstick calibration values, or information from other external sources. You may be prompted to enter data, etc. (step 266). In some cases, user-entered data may be used, for example, in recalibration routines, along with slow moving averages or quantitative differences in sensitivity, to determine new or updated sensitivities.

FIG. 25 shows another flowchart 286 of an exemplary method using slow moving averages or steady state values. In a first step, after initial calibration, analyte concentration values continue to be monitored with the sensor (step 288). Initial calibration can be based on many factors, including population averages, data from previous sessions, bench data, in vitro data, or other a priori data (step 290).

A clinical value of the analyte concentration is calculated and/or displayed based on the measured value and the initial value. An updated calibration may then be calculated based on measurements only, eg, signals from sensors only (step 294). Adjustments may be based on slow moving average changes, steady state value changes, or other criteria.

Following the updated calibration, the clinical values may be calculated and/or recalculated based on the updated calibration (step 298), and the display may then be updated (step 300) and the values already displayed. (historical) values to updated values based on the updated calibration.

Yet another implementation of the present principles is illustrated by flow chart 302 in FIG. The first step is to receive a calibration parameter, eg, sensitivity seed value, on the monitoring device (step 304). Seed values may be based on a number of factors, including user self-characterization of medical conditions, population averages, data from previous sessions, bench data, in vitro data, or other a priori data (step 306).

The monitoring device may then continue to receive sensor data and detect when the analyte concentration values are at steady state (step 308). For example, this may be done when the set of received signals over a predetermined time period is within a predetermined range or band of values. Correlation of the measured signal values, eg, in current or counts, with known or estimated steady state values may then be performed (step 310).

Following the correlating step, the monitoring device continues to receive signals from the sensors (step 312). A clinical value of the analyte concentration is calculated and displayed (step 314) based on the received signal, the known or estimated steady state value (even if the host is no longer at steady state), and the seed value.

Behavior may be detected outside of predetermined parameters, as described above in connection with FIG. 24, where the user is prompted to enter external data, e.g., fingerstick calibration values. (step 316). A recalibration may be computed and/or recalculated and subsequently displayed (step 318) based on the received signal, external data, and optionally a seed value. In some implementations, known steady-state values and/or seed values may be reset and redisplayed (step 320) based on performed calculations and recalculated historical values.

Figures 27-33 show detailed methods for determining calibration parameters, such as sensitivity and baseline, using a probabilistic approach. Certain aspects of the probabilistic approach are owned by the assignee of the present application, entitled "Advanced Calibration For Analyte Sensors," filed March 14, 2013 and incorporated herein by reference in its entirety: It is described in US patent application Ser. No. 13/827,119. In this application, which is incorporated by reference herein, including what is referred to as a "signal-based calibration algorithm," a priori calibration distribution information is modified with real-time input to determine the calibrated data points by recursive Converted to calibrated distribution information. In this way, calibration errors can be avoided, for example, when a regression results in deviant sensitivity and/or baseline values due to inappropriate estimates on reference data. In addition to the methods described in the applications incorporated by reference above, other methods of determining calibration parameters such as sensitivity and baseline may also be used. These methods include techniques that allow factory calibration that can be used during the sensor session.

In FIGS. 27-33, distributions are again used for calibration parameters such as sensitivity and baseline, but it is data that is subsequently known, e.g. Optimized based on count distribution. Referring initially to flow chart 322 of FIG. 27, the first step is to receive an initial value of analyte concentration from the sensor, as well as an initial value or distribution of sensitivity and optionally baseline (step 324). In some cases, the baseline effect may be reduced to essentially zero to simplify calculations. The initial value for sensitivity may be from the sources described above, e.g., entered by the user, drawn from population averages, transferred from previous sessions, or other sources of seed values. Good (step 326). The initial value may also be used as part of the criteria for the slow moving average filter or in calculations. If the initial value of sensitivity or baseline is the distribution of such values, it may be developed, at least initially, from demographic considerations.

Analyte signals are then monitored from the sensors (step 328). A plurality of clinical values are then calculated and displayed (step 330) based on the initial value or distribution of monitored signal and sensitivity or the initial value of mean glucose. For simplicity, the baseline is assumed to be zero or negligible. The initial value or distribution of sensitivity is such that a user-displayed analyte concentration value can be provided even if it is less accurate during this initial period than after additional data has been collected. assumed.

A value distribution of the monitored signal may then be determined (step 332). A representative value, eg, mean, median, median, etc., of the value distribution of the monitored signal may be calculated (step 334). As an initial sensitivity, the representative value may be divided by the initial value of the analyte concentration (step 338) based on the initial assumed sensitivity.

The initial value or value distribution of the sensitivity and/or the plurality of concentration values may then be optimized to match the value distribution of the monitored signal (step 336). More specifically, sensor count is the product of sensitivity and analyte concentration, so concentration equals sensor count divided by sensitivity. The median sensor count may be determined, for example, after one day's data is acquired, and given the distribution or samples from the sensitivity parameter space and the baseline parameter space, find the best fit to the distribution of sensor counts. A search may be performed that optimizes or provides. For example, if a user's long-term glucose values over a day were in the range of 100-200, certain limits can be inferred regarding the sensitivity and what the baseline can be. Therefore, the sensitivity and long-term glucose values within the distribution may be selected such that their product best optimizes the measured representative value of sensor counts. Additionally, the sensitivity and long-term glucose values may be selected such that those values are "most likely" (step 340), where "most likely" means that those values means closest to the center of the corresponding distribution of . A slow moving average may be monitored (step 342) and its changes detected and analyzed (step 344) as described above.

In other words, following day 1, there is data on the hypothesized initial mean glucose value, or sensitivity, and distribution of sensor counts. From the distribution, the median sensor count can be obtained.

Sensor count SC=f _SC (SC), which has a normal distribution.

The sensitivity equation has the form:

y = mx + b

If b=0 is assumed and the equation is further specified to the mean value,

median sensor count = m * average glucose value

We also consider both sensor count and sensitivity as normal distributions.

f _SC (SC)=m*GV

Furthermore, m is known to be a slow transit time function due to drift, so

f _SC (SC)=m(t)*GV

It is therefore clear that sensor counts and glucose values are connected by a multiplicative "constant" that is actually a slow moving time function.

The distribution of sensor counts may also reveal potential distributions of sensitivities, ie potential initial sensitivities m, and specifically the following aspects.

Average GV = (median SC/m _median )

therefore,

m _median = median SC/mean GV

For example, if the median SC is 131,000 and the mean GV is 131, then m is 1000 counts/(mg/dL). The distribution of m may be checked to determine if this value is reasonable or unlikely. A similar determination may also be made for 'b' if it is not negligible.

A representative value of the monitored signal (sensor count) may be converted into an estimate of long-term glucose.

Long Term Glucose = (Long Term Sensor Counts)/(Sampled Sensitivity) - Sampled Baseline (mg/dL)

Graph 346 of FIG. 28 shows an exemplary distribution of sample sensitivity, graph 348 from FIG. 29 shows an exemplary distribution of sample baseline, and graph 350 of FIG. 4 shows an exemplary distribution of values. As an example, if the representative value of sensor counts is 113,536, then given the three distribution constraints described above, the optimal slope is 890 counts/(mg/dL), the optimal long-term glucose A reasonable estimate is 153.5685 and the optimal baseline is -26 mg/dL. These points are shown on the same set of graphs 346, 348 and 350 and reproduced on FIGS. 31, 32 and 33 at points 354, 358 and 362, respectively.

In variations, the distributions can be made more "granular" such that different distributions can be provided for different demographic populations or groups. Other variations will also be appreciated.

As noted above, the slow moving average filter can be used as part of drift quantification as the advanced sensor is highly selective for glucose and therefore sensitivity is the main factor contributing to changes in the slow moving average. . To initially seed the slow moving average filter, the initial seed value for the average glucose concentration may be multiplied by the average sensitivity to obtain the initial number of counts. After that,
S _t = α filter _t α*S _t-1 + (1-α) S TX _t sensor _t

After a period of time, eg, one day, enough data can be received so that it can be corrected to the actual average value, which can be used for drift determination. The above steps may then be repeated to determine the subsequent best combination of slope, baseline, and glucose in the above method, e.g., raw counts determined from Day 1 data. yields the best raw count to match the estimate of .

The distribution may, in some implementations, be modified according to the measured data as more are acquired. In this way a better calibration can be obtained. At the start, only the assumed distribution was used. After that, the actual measured data is available and can be used. The filters may be reseeded with median or other representative sensor counts and the distributions generally converge to the actual measured data. Fingerstick data is available, which can be used for even faster convergence.

Seeding or reseeding may be performed in a number of ways and adjusted to allow faster convergence of the drift curve based on the filter seed parameters. For example, the initial seed value for the filter may be based on the predicted average glucose level and the predicted signal level estimated from the sensor sensitivity. The initial seed value for glucose may also be based on the subject's demographics, such as the user's age and the duration of their diabetes. Initial seed values may also be used to determine whether fasting glucose levels, glycated hemoglobin (A1C) tests, current diabetes treatment (e.g., oral medications, basal insulin use, or intensive insulin therapy), or downloaded self-monitored blood glucose values. It may be based on data such as recorded information or laboratory tests.

In another implementation, an initial seed value may be used first to start the forward-running filter. After a representative set of sensor readings has been collected, eg, 24 hours after the sensor readings, a second filter can then run in reverse. If a representative set of sensor readings is available, the forward or reverse filter can be seeded with a typical signal value, such as the median sensor reading, or the filter can be 0.9*median It can be seeded with a typical signal value that is adjusted for the expected drift, such as starting the forward filter at the value and the backward filter at 1.1*median. These techniques are then further optimized such that when more than one filter is used, such as a forward filter and an inverse filter, their seed values minimize the difference between the two filter outputs. have the advantage of obtaining For example, one exemplary method is to minimize the mean squared error between two filter outputs.

In systems and methods according to the present principles described above, daily redefinition of the seed value helps minimize the mean absolute error in the signal domain. In one implementation, using rough estimates of the signal trajectory from day 1, for example, each day, the smooth trend of the noisy filter output can create a trajectory for day 2 drift. Drift rate estimates can be compared from two or more different methods, the difference or error between the two determining the distribution of sensitivities, particularly with respect to certainty intervals, in the patent application incorporated by reference above (No. 13 /827,119) to an algorithm such as a signal-based calibration algorithm. Also, in this way, signal features may be extracted, including features corresponding to noise, level, drift, model, power, energy, and the like. In this way, it can be determined whether the drift correction is on proper start-up. For example, if the drift slope differs significantly from what the factory calibration model suggests, e.g., more than a predetermined threshold percentage, the user may be prompted for feedback or to provide a fingerstick calibration. can be prompted to

In some implementations, a smoothing filter may be used to compensate for signal drift in real time. In one case, a double exponential smoothing filter is used. Such filters may estimate multiplicative decay trends without seasonality, but may estimate additive or multiplicative seasonality to improve performance. A double exponential filter acts to recover the underlying change, or drift, of the sensor signal as a function of time. There are three main underlying equations that govern the double exponential smoothing filter.

A table of the parameters in the equations can be found in Table I below.

In the above equations and in this setting, α and β can be considered small. α is small because it is desirable not to have the filter affected by glucose excursions. β is small because the underlying trend being restored is inherently slow moving. The following parameters were used to generate one set of exemplary results (Table II (estimating sampling rate for 5 minutes)).

In the above equation, the slope may be the initial sensitivity calculated by the algorithm or a value determined from another method or from the method for determining sensitivity values described above. The average glucose in the above equation may be the average of historical glucose values from previous sessions or based on A1C values reported by the user, for example. In one implementation, data were generated using initial sensitivities estimated by the algorithm using 2-hour calibration and self-reported A1C counts from the cohort of test subjects.

The following equation was used to estimate the drift correction curve.

The sensor signal was then drift corrected using the following equation.

Glucose values were then calculated using the following equation.

In the above equation, the slope is that estimated by the algorithm at the initial calibration, and the baseline is the slope multiplied by 1 mg/dL.

To illustrate the effectiveness of the double exponential filter, FIGS. 34 and 35 show exemplary glucose traces. FIG. 34 shows a CGM trace 364 with reference and calibration values. FIG. 35 shows the raw sensor signal 366 along with the output 367 from the double exponential filter. In this case, the sensor was calibrated once using two activation values entered by the user.

An estimated drift curve for the above sensor is seen by curve 368 in FIG. As can be seen, drift curves are readily seen in both the upward and downward directions, and knowledge of the drift detected and quantified by the double exponential filter allows correction of the drift. One advantage of this implementation is that there is no need to rely on any known curve shape to correct for drift.

37-39 are further charts showing drift correction according to the above principles.

Besides using a double exponential filter, other filters can also be used. For example, a Kalman filter may be used, including process noise, also known as model noise, and measurement noise estimates. Gaussian filters, as well as traditional Butterworth low pass, moving median, moving average filters, etc. may also be used, as long as the filters help reveal underlying trends. A filter bank or series of filters may be used that combine multiple filters to obtain an average or combined trend of the underlying signal. When multiple filters are used, different types of filters may be used within a single bank, and filter settings may differ between filters. Through the use of multiple filters, signal correction may be improved at the end of subsequent time periods, eg, at the end of the second day's worth of data, at the end of the third day's worth of data, and so on. Without wishing to be bound by theory, it is assumed that in using these filters the day-to-day variability of average glucose readings is insignificant compared to slope changes or underlying signal changes.

In another variation, the signal may be preprocessed before drift estimation filtering to remove data gaps and outliers. Additionally, the calibration may be automatically updated in a manner that reduces the occurrence of unexpected spikes in CGM readings. Such methods include making the change when the signal is stable, e.g., changing the calibration setting in the middle of the night, or slowly changing the calibration to the current setting over a period of time, e.g., an hour or more. Including mixing.

Other useful techniques that can be used with filtering include various decomposition techniques. For example, an empirical model decomposition can be used to split the signal into a series of eigenmode functions over time. Other time- and frequency-based decompositions may be used, including Fourier transforms and wavelet transforms.

In another variation, other signal-based parameters can be determined and used in calibration. For example, referring to FIG. 40, it can be seen that the coefficient of variation (CV) of the sensor signal has a strong correlation with the glucose variability of the signal, eg, the glucose standard deviation. Therefore, in determining the calibration parameters, this correlation may be used to select calibration parameters that satisfy the signal CV-glucose standard deviation error model.

More specifically, a priori information as described can be used in factory calibration, which includes information obtained prior to a particular calibration. For example, such information includes information from previous calibrations, such as information obtained prior to sensor insertion. Calibration information is useful for calibrating continuous glucose sensors such as, but not limited to, sensitivity (m), change in sensitivity (dm/dt), and other signal and time derivative aspects as described above. information. Also importantly, a priori information includes distribution information such as ranges, distribution functions, and distribution parameters including mean and standard deviation.

Specifically, with respect to the standard deviation of the distribution of glucose values, it can be advantageously used, for example, in determining likely limits of glucose values and possible combinations of sensitivity and baseline. The standard deviation may also be used in determining the level of certainty from a priori calibration distribution information (such as when it is feedback of a posteriori calibration distribution information from previous calibrations (the tightness of the distribution). where the level of tightness or laxity is quantified by the standard deviation)). In the same way, the level of certainty may be determined from the a posteriori calibrated distribution information, eg, based on the level of tightness or looseness of the distribution, which again can be quantified by the standard deviation.

As a specific example, FIG. 41 shows glucose signals over 10 days. From this, the signal standard deviation and signal mean may be calculated. A signal variation coefficient may then be determined by:

Signal CV = Signal standard deviation / Signal average value

For FIG. 41, the signal CV was calculated as 0.4230.

If a relationship has been determined between the signal CV and the glucose standard deviation (see, for example, the line in FIG. 42), the calculated signal CV determined above determines the predicted glucose standard deviation. , which may then be used in the above calculations as well as other calculations. For example, an error model may be established using this correlation, and the error model is used in calibration in US patent application Ser. No. 13/827,119, incorporated above by reference.

FIG. 43 shows an exemplary distribution of differences between measured and predicted standard deviations.

Other relationships can also be used. For example, referring to FIG. 44, the relationship between mean glucose values and glucose standard deviation is seen. Specifically, it can be seen that users with higher standard deviations tend to have higher average glucose. This relationship may be used in determining, setting, inferring or otherwise selecting calibration values. As another example, and with reference to FIG. 45, another useful indicator is patient type. Specifically, FIG. 45 shows a clear difference in glucose standard deviation between non-diabetics, type I diabetics, and type II diabetics. Accordingly, based on patient type, the model selected for the patient population can be modified based on the population type, e.g., the standard deviation can be strengthened with respect to the CV error model, the mean shifted and what can be done.

As described above, the system may adjust the data for time lag from data over a previous period (e.g., to remove time lag induced by real-time filtering), and the time lag adjusted data over that period. A graphical representation may be displayed (eg, a trend graph). The system may also compensate for time lag. For example, FIG. 46 shows data points separated by a time lag, where Δ represents the individual rate of change between two adjacent points. Such time lags can be compensated for by using the following or similar equations.

glucose (t) = [drift correction sensor (t) + 5*ROC (t)]/mb (mg/dL)

For example, if the current point is within light noise, all filtered sensor counts may be used.

An exemplary calibration routine is now described, the steps of which can be seen in flowchart 400 of FIG. In a first step, the sensitivity profile versus time is characterized using a bench test that measures the sensor's response in a glucose solution over a period of one day (step 402). Since this is generally a destructive test, it can be performed on a representative set of sensors from a production lot or production line. This test may be repeated periodically or when the process changes, eg, when there are changes in raw materials or equipment. The sensitivity of each sensor may then be measured in a non-destructive bench test (step 404). The results of steps 402 and 404 are used to estimate the in vivo initial and final sensitivities of each sensor (step 406).

Note that the destructive test measures long term drift for a group of sensors, for example determining that a sensor drifts 10% over the first two days. A non-destructive bench test provides a starting point for each sensor in logarithms where, for example, the sensor of interest may have 20 starting sensitivities. Combining the two tests, it can be determined, for example, that the sensor of interest is expected to start at 20 and drift by 10%, eg, to 22.

Thus, this step maps or converts bench values to in vivo values using functions trained or optimized on well-characterized in vivo data, such as clinical trials. As another example, the sensor of interest may start at 24 and drift to 26 in vivo.

Since the electrical characteristics of the transmitter (such as gain and offset) are calibrated during manufacture (step 408) and these calibration coefficients are stored on the transmitter (step 412), the raw sensor signal is used by the algorithm can be corrected for part-to-part differences in the electronics before performing

The sensor is then packaged with a single-use transmitter (step 414). The sensor's identifier, e.g., an identification number, is read with an optical barcode and its estimated in-vivo sensitivity is retrieved from a manufacturing database and sent to the transmitter using, e.g., wireless (NFC or Bluetooth) communication. is written to (step 416).

When the transmitter first detects that it is connected to a functioning sensor, the session timer is started and the algorithm begins (step 418). The algorithm begins converting the sensor signal to glucose using the CGM model (step 420 ), with model parameters set to the previous information regarding sensor sensitivity recorded in step 416 .

When a representative set of signal data, e.g., 24 hours, is available, seed parameters for the forward and backward filters may be set using the signal median and the estimated amount of drift (step 422). . These seed values are then further optimized to minimize the mean squared error between the two signal filters (step 424).

Signal-based calibration algorithms use average values of forward and inverse filtered signals and raw sensor signals, while adjusting model parameters such as sensitivity and baseline to meet some criteria. (Step 426). Time-based input data is then used to update the algorithm. Exemplary signal-based calibration algorithms are described in the above-incorporated-by-reference patent application (No. 13/827,119), specifically [0188], namely, a Bayesian learning approach for drift estimation and correction. It is disclosed in Example 4 shown.

In the implementation of FIG. 47, the model is adjusted to meet the criterion that the mean glucose value matches the predicted diabetic mean, and another criterion that the CGM glucose variability matches the mean glucose level. Further adjustments may be made. To calculate these metrics, the algorithm has an estimated relationship between mean glucose and glucose variability.
For example, a non-diabetic patient may have a mean glucose level of 85 mg/dL and a standard deviation of 15 mg/dL. A diabetic patient may have a mean glucose of 170 mg/dL and a standard deviation of 65 mg/dL.

The optimized model parameters are used by the CGM model to convert subsequent sensor readings into glucose values that are then displayed (step 428).

Steps 424-428 are repeated when a new set of signal data is available.

A similar method is used to detect an unacceptable amount of sensor change (typically loss of sensitivity from day 7 onwards) and to stop displaying potentially inaccurate readings. may be used for

In one preferred embodiment, the analyte sensor is an implantable glucose sensor, such as described with reference to US Pat. No. 6,001,067 and US Patent Publication No. US-2005/0027463-A1. be. In another preferred embodiment, the analyte sensor is a transdermal glucose sensor, as described with reference to US Patent Publication No. US-2006/0020187-A1. In still other embodiments, the sensor comprises US Patent Publication No. US-2007/0027385-A1, co-pending US Patent Application Serial No. 11/543,396, filed October 4, 2006, March 26, 2007. in a host vessel or as described in co-pending US patent application Ser. Configured to be implanted externally. In one alternative embodiment, a continuous glucose sensor is described, for example, in Say et al. including a transcutaneous sensor as described in U.S. Pat. No. 6,565,509. In another alternative embodiment, a continuous glucose sensor is described, for example, in Bonnecaze et al. No. 6,579,690 to Say et al. including a subcutaneous sensor as described with reference to U.S. Pat. No. 6,484,046. In another alternative embodiment, a continuous glucose sensor is described, for example, in Colvin et al. including replaceable subcutaneous sensors such as those described with reference to US Pat. No. 6,512,939. In another alternative embodiment, a continuous glucose sensor is described, for example, in Schulman et al. including an intravascular sensor as described with reference to U.S. Pat. No. 6,477,395. In another alternative embodiment, the continuous glucose sensor is as described in Mastrototaro et al. including an intravascular sensor as described with reference to U.S. Pat. No. 6,424,847.

Connections between elements shown in the figures represent exemplary communication paths. Additional communication paths, either directly or through intermediates, may be included to further facilitate the exchange of information between the elements. A communication path may be a two-way communication path that allows elements to exchange information.

The various operations of the methods described above may be performed by any suitable means capable of performing the operations, such as various hardware and/or software component(s), circuits, and/or module(s). may be implemented. Generally, any operations shown in the figures may be performed by corresponding functional means capable of performing the operations.

The various exemplary logic blocks, modules, and circuits described in connection with the present disclosure (such as the blocks of FIGS. 2 and 4) are general purpose logic blocks designed to perform the functions described herein. a processor, digital signal processor (DSP), application specific integrated circuit (ASIC), field programmable gate array signal (FPGA) or other programmable logic device (PLD), separate gate or transistor logic, separate hardware component, or Any combination of these may be implemented or performed. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any commercially available processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, such as a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such combination of configurations. .

In one or more aspects, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage media may be any available media that can be accessed by a computer. By way of example, and not limitation, such computer readable media may comprise various types of RAM, ROM, CD-ROM, or other optical disk storage, magnetic disk storage, or other magnetic storage devices, or instructions or data structures. and any other medium that can be used to carry or store desired program code in the form of a computer and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, software may be transferred from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave. When transmitted, coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of medium. Disk and disc, as used herein, include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disc, and Blu-ray®. A disk typically reproduces data magnetically, while a disc reproduces data optically with a laser. Thus, in some aspects computer readable medium may comprise non-transitory computer readable medium (eg, tangible medium of expression). In addition, in some aspects computer readable medium may comprise transitory computer readable medium (eg, a signal). Combinations of the above should also be included within the scope of computer-readable media.

The methods disclosed herein comprise one or more steps or actions for achieving the described method. The method steps and/or actions may be interchanged with one another without departing from the scope of the claims. In other words, unless a specific order of steps or actions is specified, the order and use of specific steps and/or actions may be modified without departing from the scope of the claims.

Certain aspects may include computer program products for performing the operations presented herein. For example, such a computer program product may include a computer-readable medium having instructions stored (and/or encoded) thereon, the instructions for performing the operations described herein. It can be executed by one or more processors. For certain aspects, a computer program product may include packaging materials.

Software or instructions may also be transmitted over a transmission medium. For example, software may be transferred from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave. When transmitted, coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included within the definition of transmission medium.

Additionally, modules and/or other suitable means for implementing the methods and techniques described herein may be downloaded and/or otherwise obtained by user terminals and/or base stations, where applicable. should be understood. For example, such devices may be coupled to servers to facilitate transfer of means for implementing the methods described herein. Alternatively, the various methods described herein may be provided via storage means (eg, RAM, ROM, physical storage media such as compact discs (CDs) or floppy disks, etc.) such that user terminals and/or Alternatively, the base station may acquire various methods upon coupling or providing storage means to the device. In addition, any other suitable technique for providing the devices with the methods and techniques described herein may be utilized.

It is to be understood that the claims are not limited to the precise configuration and components illustrated above. Various modifications, changes, and alterations may be made in the arrangement, operation, and details of the methods and apparatus described above without departing from the scope of the claims.

Unless otherwise defined, all terms (including technical and scientific terms) have their ordinary and customary meanings given to those of ordinary skill in the art and are not so expressly defined herein. as long as it is not limited to a special or customized meaning. If the use of a particular term in describing a particular feature or aspect of this disclosure is restricted to include any particular characteristic of the feature or aspect of this disclosure to which that term relates, then that term is It should not be construed to imply redefinition herein. The terms and phrases used in this application, particularly in the appended claims, and variations thereof, are to be interpreted as non-limiting, as opposed to limiting, unless explicitly stated otherwise. As an example of the above, the term "including" should be construed to mean "including without limitation," "including, but not limited to," etc., and the term "comprising" As used herein, synonymous with "including," "containing," or "characterizing," inclusive or non-limiting, additional unlisted elements or Without excluding method steps, the term "comprising" shall be interpreted as "having at least" and the term "including" shall be interpreted as "including but not limited to" and the term "Examples" is used to provide illustrative examples of the subject matter under discussion, rather than an exhaustive or exclusive list, and includes "known," "conventional," "standard Adjectives such as," and terms of similar import should not be construed to limit the matter described to that which is available for a given period of time or at any given point in time, but rather the present or "preferred," "preferred," "desired," and "preferred" should be construed to encompass known, conventional, or standard techniques that may be available or known at any time hereafter. , or “desirable,” and terms of similar import should be understood to imply that a particular feature is critical, essential, or even important to the structure or function of the present invention. rather, it is merely intended to highlight alternative or additional features that may or may not be utilized in a particular embodiment of the invention. Similarly, a group of items joined by the conjunction "and" should not be construed to require that each one of those items be present in the group; should be construed as "and/or". Similarly, groups of items joined by the conjunctive "or" should not be construed as requiring mutual exclusivity between the groups, but rather "and/or" unless stated otherwise. should.

When a range of values is provided, it is understood that the upper and lower limits and each intervening value between the upper and lower limits of the range are included within the embodiment.

Regarding substantially any plural and/or singular terms herein, those of ordinary skill in the art can adjust plural to singular and/or vice versa as appropriate to the context and/or application. can be converted into a form The various singular/plural permutations may be expressly set forth herein for the sake of clarity. The indefinite articles "a" or "an" do not exclude plural forms. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different and independent claims does not indicate that a combination of these measures cannot be used to advantage. Reference signs in the claims shall not be construed as limiting the scope.

Where a particular number is contemplated in an introduced claim statement, such intent is expressly recited in the claim, and in the absence of such a statement, no such intent exists. will be further understood by those skilled in the art. For example, to aid understanding, the following appended claims may contain usage of the introductory phrases "at least one" and "one or more" to introduce claim recitations. However, use of such phrases may limit the use of the indefinite article even when the same claim contains the introductory phrase "one or more" or "at least one" and an indefinite article such as "a" or "an." the introduction of a claim recitation by "a" or "an" limits any particular claim containing such introduced claim recitation to embodiments containing only one such recitation; (e.g., "a" and/or "an" should typically be construed to mean "at least one" or "one or more") , as well as the use of definite articles used to introduce claim recitations. In addition, even where a particular number of an introduced claim recitation is explicitly stated, those skilled in the art will typically interpret such recitation to mean at least that stated number. (e.g., a mere statement of "two statements" without other modifiers typically means at least two statements, or two or more statements ). Further, where conventional expressions similar to "at least one of A, B, and C, etc." are used, generally such structures will be understood by those skilled in the art to is intended to include any combination of the recited items, including, for example, single members (e.g., "a system having at least one of A, B, and C" means A (including, but not limited to, systems having A alone, B alone, C alone, A and B, A and C, B and C, and/or A, B, and C, etc.). Where a conventional expression similar to "at least one of A, B, or C, etc." is used, such a structure is generally interpreted in the sense that a person skilled in the art would understand the conventional expression. (e.g., "a system having at least one of A, B, or C" includes A alone, B alone, C alone, A and B, A and C, B and C, and/or (including but not limited to systems having A, B, and C, etc.). Substantially any disjunctive words and/or phrases indicating two or more alternative terms refer to one of the terms, any of the terms, whether in the description, claims, or drawings. , or should be understood to contemplate the possibility of including both terms. For example, the phrase "A or B" is understood to include the possibilities of "A" or "B" or "A and B."

All numbers expressing quantities of ingredients, reaction conditions, etc. used herein are to be understood to be modified in all instances by the term "about." Accordingly, unless indicated to the contrary, the numerical parameters set forth herein are approximations that may vary depending on the desired properties sought to be obtained. At a minimum, and not as an attempt to limit the application of the doctrine of equivalence of any claim in any application claiming priority to this application, each numerical parameter shall have a significant number of digits and the usual should be interpreted taking into account the rounding method.

All references listed herein are hereby incorporated by reference in their entirety. To the extent any publications and patents or patent applications incorporated by reference contradict the disclosure contained herein, the present specification may supersede and/or supersede any such conflicting material. Intend.

Headings are included herein for reference and to aid in locating the various sections. These headings are not intended to limit the scope of the concepts described therein. Such concepts may have applicability throughout the specification.

Furthermore, although the foregoing has been described in some detail by way of illustration and example for purposes of clarity and understanding, it will be apparent to those skilled in the art that certain changes and modifications may be made. Therefore, the description and examples should not be construed as limiting the scope of the invention to the particular embodiments and examples described herein, but rather all that fall within the true scope and spirit of the invention. should be construed to include modifications and alternatives of

2 drug delivery pump 4 glucose meter 8 continuous analyte sensor system 10 continuous analyte sensor 12 sensor electronics 14, 16, 18, 20 display device 100 system 406 network 490 cloud-based analyte processor

Claims (129)

A method of calibrating an analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, wherein at steady state the analyte concentration value in the biological system is known, said method teeth,
a. detecting on the monitoring device when an analyte concentration value measured by an analyte concentration sensor indwelling within the biological system constitutes a first repetitive event;
b. measuring the analyte concentration value when the biological system is at the detected first repetitive event on the monitoring device or on a device or server operably coupled to the monitoring device; and correlating the analyte concentration values of . 2. The method of claim 1, wherein said correlating comprises determining a functional relationship between said sensor readings and said known analyte concentration values. 3. The method of claim 2, wherein the functional relationship comprises a multiplicative constant. A method according to any one of claims 1 to 3, wherein said detecting comprises waiting a predetermined amount of time following entry of an event into said monitoring device. 5. The method of claim 4, wherein the event is diet or exercise. a. following the correlation, detecting the occurrence of a second repetitive event, wherein the second repetitive event is different from the first repetitive event;
b. recalibrating the analyte concentration sensor by correlating the sensor reading when the biological system is at the detected second repetitive event with the known analyte concentration value. , the method according to any one of claims 1 to 5. 7. The sensor reading of claim 6, wherein the sensor reading has a first raw value at initial calibration and a second raw value at recalibration, the first and second raw values being different. the method of. The recalibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby preventing unexpected spikes in readings from occurring. 7. The method of claim 6, reduced. a. following the correlation, displaying a graph or table showing current measured and historical values of the analyte concentration calibrated at least in part based on the correlation;
b. 8. The method of claim 7, further comprising, following the recalibration, updating the display of the graph or table showing currently measured and historical values of the analyte concentration according to the recalibration. the method of. 10. The method of claim 9, wherein said updating changes said display of said historical value of said analyte concentration. a. determining a difference between the first raw value and the second raw value;
b. 8. The method of claim 7, further comprising comparing the difference-based quantity to a predetermined reference and determining whether the sensor calibration is drifting based on the comparison. 12. The method of claim 11, further comprising determining a quantitative amount by which the sensor calibration has drifted. 13. The method of claim 12, further comprising adjusting the sensor calibration based on the determined quantitative quantity. 12. The method of claim 11, wherein said quantity is a slope between said first raw value and said second raw value. 15. The method of claim 14, wherein if the slope exceeds a predetermined threshold, future calibrations are prohibited until the slope no longer exceeds a predetermined threshold. 16. The method of claim 15, further comprising prompting a user to enter measurements. The method of any one of claims 1-11, wherein the sensor is a glucose sensor. a. receiving a signal from the sensor subsequent to the correlating;
b. displaying a value corresponding to the received signal, wherein the displayed value is based on the received signal and the known analyte concentration value. A method according to any one of 19. The method of any one of claims 1-18, further comprising determining the known analyte concentration value by prompting a user to enter a measurement value. 20. The method of any one of claims 1-19, further comprising determining the known analyte concentration value by accessing a population average. 21. The method of any one of claims 1-20, wherein the repetitive event is selected from the group consisting of steady state, postprandial rise, daily high and low glucose range, decay rate, or rate of change. A method for compensating for drift in an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising:
a. measuring an analyte value using an indwelling analyte concentration sensor;
b. determining a first slow moving average of the analyte measurements over a first time period, and referencing a calibration of the sensor based at least in part on the first slow moving average;
c. following said first determination, determining a second slow moving average of said measurements of said analyte over a second time period;
d. adjusting the calibration of the sensor based at least in part on the difference between the first slow moving average and the second slow moving average. 23. The method of claim 22, wherein the duration of said first time period is greater than about 12 hours. 24. The method of claim 23, wherein the duration of said first time period is greater than about 24 hours. A method according to any one of claims 22 to 24, wherein the duration of said first period is the same as the duration of said second period. a. at least a history of calibrated analyte concentrations based at least in part on said first slow moving average subsequent to said referencing said calibration of said sensor based at least in part on said first slow moving average; displaying a graph or table showing the values;
b. following said adjustment, updating said display of said graph or table showing at least historical values of said analyte concentration according to said adjusted calibration. The method described in . 27. The method of claim 26, wherein said updating changes said display of said historical value of said analyte concentration. 27. The method of claim 26, wherein the displayed graph or table further indicates the currently measured value of the analyte concentration. 28. The method of any one of claims 22-27, wherein said referencing a calibration of said sensor based at least in part on said first slow moving average further comprises basing said calibration on a seed value. the method of. 30. The method of claim 29, wherein the seed value is a value received from a population average or from a previous session. 31. The method of claim 30, further comprising, following said adjustment, varying said seed value based at least in part on said adjustment. 32. The method of claim 31, further comprising varying the seed value based on a difference between the first slow moving average and the second slow moving average. Said adjustment of said calibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby preventing unexpected spikes in readings. The method of any one of claims 22-32, wherein incidence is reduced. A method for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising:
a. measuring an analyte value using an indwelling analyte concentration sensor;
b. determining a first slow moving average of the analyte measurements over a first set of time periods, the first set including an event-based time period; at least in part typically referencing calibration of the sensor based on the first slow moving average;
c. determining, following the first determination, a second slow moving average of the measurements of the analyte over a second set of time periods, the second set of event-based time periods; including determining and
d. adjusting the calibration of the sensor based at least in part on the difference between the first slow moving average and the second slow moving average. 35. The method of claim 34, wherein the first and second event-based time periods are selected from the group consisting of a postprandial period, a sleep period, and a postbreakfast period. Said adjustment of said calibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby preventing unexpected spikes in readings. 35. The method of claim 34, wherein incidence is reduced. A method for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising:
a. measuring an analyte value using an indwelling analyte concentration sensor;
b. determining a first slow moving average of measurements of the analyte over a first time period and referencing a first apparent sensitivity of the sensor based at least in part on the first slow moving average;
c. determining a second slow moving average of the measurements of the analyte over a second time period following the first determination; and determining a second slow moving average of the sensor based at least in part on the second slow moving average based on the apparent sensitivity of
d. determining if the change in apparent sensitivity of the sensor between the first apparent sensitivity and the second apparent sensitivity meets a predetermined criterion;
e. adjusting the actual sensitivity of the sensor to an adjusted value based on the difference between the first apparent sensitivity and the second apparent sensitivity if the change in apparent sensitivity meets a predetermined criterion; When,
f. prompting a user to enter data if said change in apparent sensitivity does not meet predetermined criteria so that a reason for said change in apparent sensitivity can be determined. 38. The method of claim 37, wherein said determining comprises determining whether said change in apparent sensitivity is due to sensitivity drift or change in said slow moving average. 39. The method of claim 38, further comprising, if the change in apparent sensitivity is due to a change in the slow moving average, prompting the user to enter data related to the change. 38. The method of claim 37, wherein the predetermined criteria include known behavior of sensitivity changes over time for the sensor. 41. The method of claim 40, wherein the known behavior of sensitivity change constitutes an envelope of acceptable sensitivity change over time. 38. The method of claim 37, wherein said adjusted value is based at least in part on said second slow moving average. 41. The method of claim 40, wherein said predetermined criteria further comprise known values for physiologically likely analyte changes. 38. The method of claim 37, wherein the prompting the user includes prompting the user to enter calibration values. 38. The method of claim 37, wherein the prompting the user comprises prompting the user to enter diet or exercise information. 3. Upon receiving said calibration values or said diet or exercise information from said user, said method further comprises determining whether said change in apparent sensitivity is due to sensitivity drift or change in said slow moving average. 46. The method of any one of 44 or 45. 47. The method of claim 46, further comprising adjusting the actual sensitivity of the sensor based on the received calibration value or dietary or exercise information. The adjustment of the actual sensitivity of the sensor is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings over a threshold period of time, whereby the readings 48. The method of any one of claims 37-47, wherein the occurrence of unexpected spikes in is reduced. A method for checking calibration of an analyte concentration sensor system in a biological system using only signals from the analyte concentration sensor, comprising:
a. After initial calibration, measuring analyte values over time using an indwelling analyte concentration sensor;
b. calculating a clinical value of analyte concentration based on the measurements and the initial calibration;
c. adjusting an initial calibration to an updated calibration, the adjustment being based only on said measurements of said analyte over time or a subset thereof;
d. calculating a clinical value for the analyte concentration based on the measured value and the updated calibration. 50. The method of claim 49, wherein the initial calibration is based on population averages or user-entered data. 51. The method of claim 49 or 50, wherein said adjustment is based on a slow moving average of said measurements of said analyte over time. 52. The method of any one of claims 49-51, wherein said adjustment is based on the steady state value of said analyte. 53. The method of any one of claims 49-52, wherein said initial calibration is based on data determined prior to a session associated with said indwelling analyte concentration sensor. 54. The method of claim 53, wherein the data are determined a priori, on the bench, or in vitro. The adjustment is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby reducing the occurrence of unexpected spikes in readings. The method of any one of claims 49-54, wherein A method of calibrating an analyte concentration sensor in a biological system using a signal from an analyte concentration sensor, wherein at steady state an analyte concentration value in the biological system is known, the method comprising: ,
a. Receiving seed values for calibration parameters on a monitoring device;
b. detecting on the monitoring device when an analyte concentration value measured by an analyte concentration sensor indwelling within a biological system is at steady state;
c. measuring the analyte concentration value when the biological system is at the detected steady state on the monitoring device or on a device or server operably coupled to the monitoring device; correlating with a value;
d. receiving a signal from the sensor subsequent to the correlating;
e. calculating and displaying a value corresponding to the received signal, the calculated value being based on the received signal, the known analyte concentration value, and the seed value; A method, including 57. The method of claim 56, wherein the received seed value is received from a source containing factory calibration information. a. detecting behavior in the received signal other than predetermined parameters;
b. 58. The method of claim 56 or 57, further comprising prompting the user to enter external calibration information. 59. The method of Claim 58, wherein the displayed value is further based on the external calibration information. 59. The method of claim 58, wherein the external calibration information is received from SMBG or fingerstick calibration. 61. The method of any one of claims 58-60, further comprising resetting the known calibration values to new known calibration values, the resetting being based at least in part on the external calibration information. 61. The method of any one of claims 58-60, further comprising resetting the seed value to a new seed value, the reset being based at least in part on the external calibration information. 62. The method of any one of claims 56-61, further comprising modifying the display based on the determined accuracy of the value. 64. The method of claim 63, wherein said altering said display includes displaying ranges rather than values or vice versa. 65. The method of any of claims 56-64, wherein the received seed values of calibration parameters are user-entered medical condition characteristics. 66. The method of claim 65, wherein the user-entered medical condition characteristics include an indication of type I diabetes, type II diabetes, non-diabetes, or pre-diabetes. 67. The method of any one of claims 56-66, wherein the received seed values for calibration parameters are values based on one or more user-entered blood glucose levels. wherein said displaying values corresponding to said received signals includes displaying a graph or table showing currently measured and historical values of said analyte concentration;
a. detecting that a calibration change has occurred;
b. adjusting one or more calibration parameters of the analyte concentration sensor according to the change in calibration;
c. 57. Following said adjustment, updating said display of said graph or table showing currently measured and historical values of said analyte concentration according to said adjusted calibration parameters. 67. The method of any one of paragraphs 1-67. said detecting that a change in calibration has occurred,
a. detecting slow moving average changes, or b. 69. The method of claim 68, comprising detecting changes in the steady state value. The adjustment is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby reducing the occurrence of unexpected spikes in readings. The method of any one of claims 56-69, wherein A method of calibrating an analyte concentration sensor, wherein following sensor insertion into a patient only parameters based on or derivable from the sensor signal are used,
a. receiving at least an initial value of analyte concentration and an initial value or initial value distribution of sensor sensitivity;
b. following insertion of an analyte concentration sensor, monitoring the signal from said sensor for a duration of time;
c. calculating a plurality of analyte concentration values based on the monitored sensor signal and the initial value or value distribution of the sensor sensitivity over the duration;
d. determining a value distribution of the monitored signal over the duration;
e. optimizing the initial value or the value distribution of the sensor sensitivity and the plurality of analyte concentration values to match the value distribution of the monitored signal;
f. determining an updated sensitivity based on said optimization. 3. The receiving is receiving an initial distribution of sensor sensitivities, and wherein the calculating a plurality of analyte concentration values is based on representative values from the monitored sensor signal and the initial distribution of sensor sensitivities. 71. The method according to 71. 73. The method of claim 72, wherein said representative value is selected from mean or midpoint or median. Said determination of updated sensitivity comprises:
a. dividing the representative value by the initial value of the analyte concentration;
b. 74. The method of claim 73, further comprising updating the sensitivity value to equal the result of the division. 75. The method of any one of claims 71-74, wherein the initial value of the analyte population is the population mean. 76. The method of any one of claims 71-75, wherein the initial value of the analyte population is entered by a user. A method according to any one of claims 71 to 76, wherein said initial value of analyte population is transferred from a previous session. 78. The method of any one of claims 71-77, wherein said optimizing comprises optimizing the product of said initial value or value distribution of said sensor sensitivity and said plurality of analyte concentration values. The optimization of the product optimizes the product to match the value distribution of the monitored signal while corresponding parameters of the value distribution of the sensor sensitivity and a plurality of analyte concentration values. 79. The method of claim 78, comprising adjusting to most closely match the population mean. The receiving further includes receiving an initial value distribution of a baseline, and the optimizing includes combining the value distribution of the baseline with the value distribution of the sensor sensitivity and the plurality of analyte concentration values into the 80. The method of any one of claims 63-79, further comprising optimizing to match the value distribution of the monitored signal. 81. The method of claim 80, wherein the initial value distribution of baseline follows a normal distribution. 82. The method of any one of claims 63-81, wherein at least said initial value of said analyte concentration is used as part of a seed value input to a slow moving average filter. A method according to any one of claims 71 to 82, wherein said initial value distribution of sensor sensitivity is defined by a normal distribution. A method according to any one of claims 71 to 83, wherein said determined value distribution of said monitored signal follows a lognormal distribution. 85. The method of any one of claims 63-84, wherein the duration is one day. 86. The method of any one of claims 63-85, further comprising continuing to determine updated sensitivity based on previous updated sensitivity and received analyte concentration values. 83. The method of claim 82, further comprising detecting a slow moving average of said monitored analyte concentration values. 88. The method of claim 87, prompting a user to enter data if the absolute value of change in the slow moving average is greater than a predetermined threshold over a predetermined unit time. If the absolute value of change in the slow moving average is greater than a predetermined threshold over a predetermined unit time, determining whether the change is due to a system error or a change in the actual sensitivity of the sensor. The method according to 87. wherein said determining whether said change is due to system error or due to change in actual sensitivity of said sensor comprises determining whether subsequent behavior of said sensitivity is consistent with a known sensitivity profile. 90. The method of Paragraph 89. 91. The method of claim 90, wherein the known sensitivity profile is the envelope of a sensitivity curve. If it is determined that the absolute value of change in the slow moving average is due to a change in the actual sensitivity of the sensor, updating the sensitivity based at least in part on the value of the change in the slow moving average. 92. The method of claim 91, wherein The determination whether the change is due to system error or due to changes in the actual sensitivity of the sensor determines whether the subsequent behavior of the analyte concentration values is consistent with a known envelope of physiological feasibility. 90. The method of claim 89, comprising determining . 92. The method of claim 91, wherein if the absolute value of change in the slow moving average is determined to be due to a system error, prompting the user to enter data. A method for calibrating an analyte concentration sensor in a biological system using a signal from the analyte concentration sensor, comprising:
a. receiving or determining seed values for calibration parameters associated with the analyte concentration sensor;
b. using the seed value to at least partially determine a calibration of the analyte concentration sensor;
c. measuring an analyte concentration value using the analyte concentration sensor;
d. displaying calibrated measurements at least partially using the seed value. 96. The method of Claim 95, wherein said receiving or determining is performed on a monitoring device in operable signal communication with said analyte concentration sensor. 97. The method of claim 96, wherein said displaying is performed on said monitoring device or on a mobile device in signal communication with said monitoring device. said displaying of said measurements comprises displaying a graph or table showing at least historical values of said analyte concentration;
a. detecting that a calibration change has occurred;
b. adjusting one or more calibration parameters of the analyte concentration sensor according to the detected change in calibration;
c. following said adjustment, updating said display of said graph or table showing at least historical values of said analyte concentration according to said adjusted calibration parameter. The method described in section. Said adjustment of said calibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby preventing unexpected spikes in readings. 99. The method of claim 98, wherein incidence is reduced. 99. The method of Claim 98, wherein said updating changes said display of said historical value of said analyte concentration. 101. The method of any one of claims 95-100, wherein the seed value is at least partially code-based. 102. The method of claim 101, wherein the code is entered into a monitoring device by a user. 102. The method of Claim 101, wherein a monitoring device is configured to receive the code without substantial involvement of the user. 101. The method of any one of claims 95-100, wherein the seed value is based at least in part on impedance measurements. 101. The method of any one of claims 95-100, wherein the seed value is based at least in part on information associated with a production lot of the sensor. 101. The method of any one of claims 95-100, wherein the seed value is based at least in part on a population average. 101. The method of any one of claims 95-100, wherein the seed value is based at least in part on the user's most recent analyte value. A method of calibrating and compensating for drift of an indwelling analyte concentration sensor in a biological system using only a signal from the analyte concentration sensor, wherein at steady state, analyte concentration values in the biological system are known. Yes, the method comprising:
a. detecting on the monitoring device when the analyte concentration value measured by the analyte concentration sensor indwelling within the biological system is at steady state;
b. measuring the analyte concentration value when the biological system is at the detected steady state on the monitoring device or on a device or server operably coupled to the monitoring device; correlating with a value;
c. determining a first slow moving average of the measurements of the analyte over a first time period, and referencing a calibration of the sensor based at least in part on the first slow moving average and the known analyte concentration value; and
d. following said first determination, determining a second slow moving average of said measurements of said analyte over a second time period;
e. adjusting the calibration of the sensor based at least in part on the difference between the first slow moving average and the second slow moving average. Said adjustment of said calibration is configured to occur when the sensor reading is substantially stable or within a predetermined range of readings for a threshold period of time, thereby preventing unexpected spikes in readings. 109. The method of claim 108, wherein incidence is reduced. A method of calibrating a first portion of a number of sensors, the second portion being in service, comprising:
a. receiving calibration data from some of the second portions of sensors;
b. and updating one or more calibration parameters of the first portion based on the received data. 111. The method of claim 110, wherein said updating is performed before said first portion is installed to a user. 111. The method of claim 110, wherein said updating is performed after said first portion is installed in a user. 113. The method of claim 112, wherein said updating is performed by transmitting new or updated calibration parameters over a network to a monitoring device or sensor electronics module associated with said sensor. 114. The method of any one of claims 110-113, wherein the second portion of the sensor is configured to be calibrated using an a priori calibration. 115. The method of any one of claims 110-114, wherein the second portion of the sensor is configured to be calibrated using user data. 116. The method of any one of claims 110-115, wherein the second portion of the sensor is configured to be calibrated using an ex vivo bench calibration. 117. The method of any one of claims 110-116, wherein the second portion of the sensor is configured to be calibrated using blood measurements. A method for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising:
a. measuring the value of the analyte as a function of time using an indwelling analyte concentration sensor;
b. filtering the measurements using a double exponential smoothing filter;
c. and, following said filtering, displaying said filtered measurements against time. The double exponential smoothing filter is

is governed by the formula,

and in the formula,

119. The method of claim 118, wherein The glucose signal as a function of time is

, where

, where

120. The method of claim 119, wherein A method for calibrating an analyte concentration sensor in a biological system using only a signal from said analyte concentration sensor, comprising:
a. receiving data corresponding to analyte concentration values over a period of time;
b. determining a coefficient of variation of the received data;
c. determining a standard deviation of the analyte concentration values based at least in part on the determined coefficient of variation;
d. using the determined standard deviation in the calculation of a calibration parameter, the calibration parameter being associated with the analyte concentration value. 122. The method of claim 121, wherein said determining a coefficient of variation of said received data comprises dividing a standard deviation of said received data by an average of said received data. A method for compensating for drift of an analyte concentration sensor in a biological system using only the signal from the analyte concentration sensor, comprising:
a. measuring an analyte value using an indwelling analyte concentration sensor;
b. determining a first slow moving average of the measurements of the analyte over a first time period and referencing a calibration of the sensor based at least in part on the first slow moving average;
c. following said first determination, determining a second slow moving average of said measurements of said analyte over a second time period;
d. adjusting the calibration of the sensor based at least in part on a seed value and a difference between the first slow moving average and the second slow moving average. 123. The method of claim 122, wherein the seed value is determined using a signal median value, a drift value, or both. 125. Any one of claims 122-124, wherein both a forward filter and an inverse filter are used, further comprising optimizing the seed value to minimize the mean squared error between the two signal filters. The method described in . further comprising adjusting the sensitivity and baseline for the sensor according to a signal-based calibration algorithm, wherein the signal-based calibration algorithm uses averages of signals from the forward and reverse filters along with the raw sensor signal. 126. The method of claim 125, wherein 127. The method of claim 126, further comprising adjusting said sensitivity and said baseline based on one or more criteria. 128. The method of claim 127, wherein the criteria include that mean glucose values should match predicted diabetes mean values. 128. The method of claim 127, wherein said criteria include that CGM glucose variability should match mean glucose levels.

Images